CN112417636A

CN112417636A - Computer-implemented method and system for autonomous vehicle operation

Info

Publication number: CN112417636A
Application number: CN202010858470.7A
Authority: CN
Inventors: A·森西; K·斯卢茨基; A·阿莎·德维; 宣蔡哲; 陈志良
Original assignee: Motional AD LLC
Current assignee: Motional AD LLC
Priority date: 2019-08-23
Filing date: 2020-08-24
Publication date: 2021-02-26
Also published as: GB2609759A; GB2594900A; KR102542970B1; GB2594900B; GB202311621D0; KR102431324B1; US11945440B2; KR20230085127A; KR20210023765A; GB2609759B; DE102020210499A1; GB2589687B; GB202112060D0; GB202211535D0; GB2589687A; KR20220112742A; US20210053569A1; GB202013093D0

Abstract

The present invention relates to computer-implemented methods and systems for autonomous operation of vehicles, providing techniques for using human driving behavior to assist in autonomous vehicle decision-making when the autonomous vehicle encounters various scenes on a roadway. For each scenario, a model may be generated based on human driving behavior that controls how the autonomous vehicle maneuvers in that scenario. As a result of using these models, the reliability and safety of autonomous vehicles may be improved. In addition, because the model is programmed into the autonomous vehicle, in many cases, the autonomous vehicle need not expend resources to implement complex calculations to determine driving behavior in real-time.

Description

Computer-implemented method and system for autonomous vehicle operation

Technical Field

This specification relates to data-driven rule manuals.

Background

Autonomous vehicles have benefits over human-driven vehicles, such as reducing the number of road traffic accidents, traffic congestion, parking congestion, and accounting for fuel efficiency. In making driving decisions, typical autonomous vehicle systems allow for the autonomous vehicle system to know objects (such as other vehicles and obstacles, etc.) in the vehicle's environment, surrounding environment, by utilizing sensor data. However, autonomous vehicle systems may benefit from using data related to human driving behavior.

Disclosure of Invention

In general, the present disclosure provides techniques for assisting an autonomous vehicle's decision making when the autonomous vehicle encounters various scenes on a roadway using human driving behavior. More specifically, the present disclosure discusses, among other things, embodiments for modeling human behavior while driving through an uncontrolled transportation hub, while driving beyond a parked vehicle, and while driving in the presence of a person crossing a road.

These and other aspects, features and embodiments may be expressed as methods, apparatus, systems, components, program products, means or steps for performing functions, and in other ways, and will become apparent from the following description, including the claims.

Drawings

Fig. 1 illustrates an example of an autonomous vehicle having autonomous capabilities.

FIG. 2 illustrates an example "cloud" computing environment.

FIG. 3 illustrates a computer system.

Fig. 4 illustrates an example architecture of an autonomous vehicle.

FIG. 5 shows an example of inputs and outputs that may be used by the perception module.

FIG. 6 shows an example of a LiDAR system.

FIG. 7 shows the LiDAR system in operation.

FIG. 8 shows additional details of the operation of a LiDAR system.

FIG. 9 shows a block diagram of the relationship between inputs and outputs of a planning module.

Fig. 10 shows a directed graph used in path planning.

FIG. 11 shows a block diagram of the inputs and outputs of the control module.

FIG. 12 shows a block diagram of the inputs, outputs and components of the controller.

Fig. 13 illustrates an example of an autonomous vehicle attempting merge (merge) at an uncontrolled traffic intersection.

FIG. 14 illustrates actions that a computer system may take in order to generate an autonomous vehicle driving model to drive through an uncontrolled traffic intersection.

Fig. 15 shows an example of an autonomous vehicle attempting to pass a stationary vehicle in the same lane.

FIG. 16 illustrates actions that a computer system may take in order to generate an autonomous vehicle driving model to drive through stationary vehicles in the same lane.

Fig. 17 shows a scenario for avoiding a person crossing a road when an autonomous vehicle is driving on a road that includes two lanes (where adjacent lanes have traffic in the same direction).

Fig. 18 shows a scenario for avoiding a person crossing a road when an autonomous vehicle is driving on a road that includes two lanes with adjacent lanes having traffic in opposite directions.

Fig. 19 illustrates a scenario for avoiding a person crossing a road when an autonomous vehicle is driving on a road with a single lane.

Fig. 20 illustrates actions that a computer system may take in order to generate an autonomous vehicle driving model to avoid an entity (e.g., avoid a crossroad human, animal, rider, or another suitable entity).

FIG. 21 illustrates actions that an autonomous vehicle may take in order to use an autonomous vehicle driving model to drive through an uncontrolled traffic intersection.

FIG. 22 illustrates actions that an autonomous vehicle may take in order to use an autonomous vehicle driving model to drive beyond a stationary vehicle in the same lane.

Fig. 23 illustrates actions that an autonomous vehicle may take in order to use an autonomous vehicle driving model to avoid an entity (e.g., a pedestrian) in the same lane.

FIG. 24 illustrates actions that a computer system may take in order to generate a model of human driving behavior for a particular traffic maneuver.

FIG. 25 shows one possible data set describing the driving behavior of a plurality of manually operated vehicles.

Detailed Description

In the following description, for purposes of explanation, numerous specific details are set forth in order to provide a thorough understanding of the present invention. It may be evident, however, that the present invention may be practiced without these specific details. In other instances, well-known structures and devices are shown in block diagram form in order to avoid unnecessarily obscuring the present invention.

In the drawings, for purposes of explanation, specific arrangements or sequences of illustrative elements, such as those representing devices, modules, instruction blocks, and data elements, are shown. However, those skilled in the art will appreciate that the particular ordering or arrangement of the illustrative elements in the drawings is not meant to require a particular order or sequence of processing or separation of processing steps. Moreover, the inclusion of illustrative elements in the figures does not imply that such elements are required in all embodiments, nor that the features represented by such elements are necessarily included or combined with other elements in some embodiments.

Further, in the drawings, connecting elements, such as solid or dashed lines or arrows, are used to illustrate a connection, relationship or association between two or more other schematic elements, and the absence of any such connecting elements does not imply that a connection, relationship or association cannot exist. In other words, some connections, relationships, or associations between elements are not shown in the drawings so as not to obscure the invention. Moreover, for ease of explanation, a single connected element is used to represent multiple connections, relationships, or associations between elements. For example, if a connected element represents communication of signals, data, or instructions, those skilled in the art will appreciate that the element represents one or more signal paths (e.g., buses) that may be required to affect the communication.

Reference will now be made in detail to embodiments, examples of which are illustrated in the accompanying drawings. In the following detailed description, numerous specific details are set forth in order to provide a thorough understanding of the various embodiments described. It will be apparent, however, to one skilled in the art that the various embodiments described may be practiced without these specific details. In other instances, well-known methods, procedures, components, circuits, and networks have not been described in detail so as not to unnecessarily obscure aspects of the embodiments.

Several of the features described below can be used independently of one another or with any combination of other features. However, any individual feature may not solve any of the problems described above, or may only solve one of the problems described above. Some of the problems discussed above may not be adequately addressed by any of the features described herein. Although headings are provided, information related to a heading but not found in the heading section may be found elsewhere in this description. The examples are described herein according to the following summary:

1. general overview

2. Overview of the System

3. Autonomous vehicle architecture

4. Autonomous vehicle input

5. Autonomous vehicle planning

6. Autonomous vehicle control

7. Modeling human driving behavior through uncontrolled traffic intersections

8. Modeling human driving behavior to drive beyond stationary vehicles

9. Modeling human driving behavior to avoid crossing roads

10. Using a model of human driving behavior in an autonomous vehicle

11. Generating models of human driving behavior using neural networks

General overview

In general, the present disclosure provides techniques for using human driving behavior to assist an autonomous vehicle's decisions when the autonomous vehicle encounters various scenes on a roadway. For each scenario, a model may be generated based on human driving behavior that controls how the autonomous vehicle maneuvers in that scenario. As a result of using these models, the reliability and safety of autonomous vehicles may be improved. In addition, because the model is programmed into the autonomous vehicle, in many cases, the autonomous vehicle need not expend resources to implement complex calculations to determine driving behavior in real-time.

Overview of the System

Fig. 1 shows an example of an autonomous vehicle 100 with autonomous capabilities.

As used herein, the term "autonomous capability" refers to a function, feature, or facility that enables a vehicle to operate partially or fully without real-time human intervention, including, but not limited to, fully autonomous vehicles, highly autonomous vehicles, partially autonomous vehicles, and conditional autonomous vehicles.

As used herein, an Autonomous Vehicle (AV) is a vehicle with autonomous capabilities.

As used herein, "vehicle" includes a means of transportation for cargo or personnel. Such as cars, buses, trains, airplanes, drones, trucks, boats, ships, submarines, airships, etc. An unmanned car is an example of a vehicle.

As used herein, a "trajectory" refers to a path or route that navigates from a first spatiotemporal location to a second spatiotemporal location. In an embodiment, the first spatiotemporal location is referred to as an initial location or a starting location and the second spatiotemporal location is referred to as a destination, a final location, a target location, or a target location. In some examples, a track is composed of one or more road segments (e.g., segments of a road), and each road segment is composed of one or more blocks (e.g., a portion of a lane or intersection). In an embodiment, the spatiotemporal locations correspond to real-world locations. For example, the space-time location is a boarding or alighting location to allow people or cargo to board or disembark.

As used herein, a "sensor(s)" includes one or more hardware components for detecting information related to the environment surrounding the sensor. Some hardware components may include sensing components (e.g., image sensors, biometric sensors), transmitting and/or receiving components (e.g., laser or radio frequency wave transmitters and receivers), electronic components (e.g., analog-to-digital converters), data storage devices (e.g., RAM and/or non-volatile memory), software or firmware components and data processing components (e.g., application specific integrated circuits), microprocessors and/or microcontrollers.

As used herein, a "scenario description" is a data structure (e.g., a list) or data stream that includes one or more classified or tagged objects detected by one or more sensors on an AV vehicle, or one or more classified or tagged objects provided by a source external to the AV.

As used herein, a "roadway" is a physical area that can be traversed by a vehicle and may correspond to a named corridor (e.g., a city street, an interstate highway, etc.) or may correspond to an unnamed corridor (e.g., a lane of travel within a house or office building, a segment of a parking lot, a segment of an empty parking lot, a dirt passageway in a rural area, etc.). Because some vehicles (e.g., four-wheel drive trucks, off-road vehicles (SUVs), etc.) are able to traverse a variety of physical areas not particularly suited for vehicle travel, a "road" may be any physical area that is not formally defined as a passageway by municipalities or other government or administrative authorities.

As used herein, a "lane" is a portion of a roadway that can be traversed by a vehicle. Lanes are sometimes identified based on lane markings. For example, the lanes may correspond to most or all of the space between the lane markings, or only a portion of the space between the lane markings (e.g., less than 50%). For example, a roadway with lane markings far apart may accommodate two or more vehicles such that one vehicle may pass another without crossing the lane markings, and thus may be interpreted as a lane narrower than the space between the lane markings, or two lanes between lanes. In the absence of lane markings, the lane may also be interpreted. For example, lanes may be defined based on physical characteristics of the environment (e.g., rocks in rural areas and trees along thoroughfares, or natural obstacles to be avoided, for example, in less developed areas). The lane may also be interpreted independently of lane markings or physical features. For example, a lane may be interpreted based on an arbitrary path in an area without obstacles that would otherwise lack features that would be interpreted as lane boundaries. In an example scenario, the AV may interpret a lane crossing an unobstructed portion of a field or open space. In another example scenario, the AV may interpret lanes through a wide (e.g., sufficient two or more lane widths) road without lane markings. In this scenario, the AV may communicate lane related information to other AVs so that other AVs may coordinate path planning between the AVs using the same lane information.

The term "over-the-air (OTA) client" includes any AV, or any electronic device (e.g., computer, controller, IoT device, Electronic Control Unit (ECU)) embedded in, coupled to, or in communication with the AV.

The term "over-the-air (OTA) update" means any update, change, deletion, or addition to software, firmware, data or configuration settings, or any combination thereof, delivered to an OTA client using a proprietary and/or standardized wireless communication technology including, but not limited to: cellular mobile communications (e.g., 2G, 3G, 4G, 5G), wireless local area networks (e.g., WiFi), and/or satellite internet.

The term "edge node" refers to one or more edge devices coupled to a network that provides a portal for communicating with the AV and may communicate with other edge nodes and cloud-based computing platforms to schedule and deliver OTA updates to OTA clients.

The term "edge device" refers to a device that implements an edge node and provides a physical wireless Access Point (AP) to an enterprise or service provider (e.g., VERIZON, AT & T) core network. Examples of edge devices include, but are not limited to: computers, controllers, transmitters, routers, routing switches, Integrated Access Devices (IADs), multiplexers, Metropolitan Area Networks (MANs), and Wide Area Network (WAN) access devices.

"one or more" includes a function performed by one element, a function performed by multiple elements, e.g., in a distributed fashion, several functions performed by one element, several functions performed by several elements, or any combination thereof.

It will also be understood that, although the terms first, second, etc. may be used in some instances to describe various elements, these elements should not be limited by these terms. These terms are only used to distinguish one element from another. For example, a first contact may be referred to as a second contact, and likewise, a second contact may be referred to as a first contact, without departing from the scope of the various described embodiments. The first contact and the second contact are both contacts, but they are not the same contact.

The terminology used in the description of the various embodiments described herein is for the purpose of describing particular embodiments only and is not intended to be limiting. As used in the description of the various embodiments described and the appended claims, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It is also understood that the term "and/or" as used herein refers to and includes any and all possible combinations of one or more of the associated manifest items. It will be further understood that the terms "comprises," "comprising," "includes" and/or "including," when used in this specification, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof.

As used herein, the term "if" is alternatively understood as being in that case, at that time, or in response to detection, or in response to determination, depending on the context. Likewise, the phrase "if it is determined" or "if [ the condition or event ] has been detected" may be understood as "upon determining" or "in response to determining" or "upon detecting [ the condition or event ] or" in response to detecting [ the condition or event ] ", depending on the context.

As used herein, an AV system refers to AV and real-time generated hardware, software, stored data, and real-time generated data that support AV operations. In an embodiment, the AV system is incorporated within the AV. In an embodiment, the AV system is distributed across multiple sites. For example, some software of the AV system is implemented in a cloud computing environment similar to the cloud computing environment 300 described below in connection with fig. 3.

In general, this document describes techniques applicable to any vehicle having one or more autonomous capabilities, including fully autonomous vehicles, highly autonomous vehicles, and conditional autonomous vehicles, such as so-called class 5, class 4, and class 3 vehicles (see SAE International Standard J3016: Classification and definition of terms related to automotive autopilot systems on roadways, which is incorporated by reference herein in its entirety for more detailed information on the level of autonomy of the vehicle). The techniques described in this description are also applicable to partly autonomous vehicles and driver-assisted vehicles, such as so-called class 2 and class 1 vehicles (see SAE international standard J3016: classification and definition of terms relating to automotive autonomous systems on roads). In embodiments, one or more of the class 1, class 2, class 3, class 4, and class 5 vehicle systems may automatically perform certain vehicle operations (e.g., steering, braking, and map usage) under certain operating conditions based on processing of sensor inputs. The techniques described in this document may benefit vehicles at all levels, from fully autonomous vehicles to vehicles operated by humans.

Autonomous vehicles offer advantages over vehicles that require a human driver. One advantage is safety. For example, in 2016, 600 million car accidents, 240 million people injured, 40000 people dead, and 1300 million vehicle collisions experienced in the united states, with an estimated social cost of $ 9100 million. From 1965 to 2015, the number of U.S. traffic accident deaths per 1 million miles driven has decreased from about 6 to about 1, due in part to additional safety measures deployed in the vehicle. For example, it is believed that the additional half second warning of a collision alleviates a 60% front-to-back collision. However, passive safety features (e.g., seat belts, airbags) may have reached their limits in improving this number. Thus, active safety measures such as automatic control of the vehicle are a possible next step to improve these statistics. It is believed that human drivers are responsible for serious pre-crash events in 95% of crashes, and therefore it is possible for an autonomous driving system to achieve better safety results, for example by: emergency situations are recognized and avoided more reliably than humans; making better decisions than humans, better complying with traffic regulations than humans, and better predicting future events than humans; and to control vehicles more reliably than humans.

Referring to fig. 1, the AV system 120 runs the AV100 along a trajectory 198, through the environment 190 to a destination 199 (sometimes referred to as a final location), while avoiding objects (e.g., natural obstacles 191, vehicles 193, pedestrians 192, riders, and other obstacles) and complying with road rules (e.g., operational rules or driving preferences).

In an embodiment, the AV system 120 comprises means 101 for receiving and operating an operation command from the computer processor 146. We use the term "operating command" to mean an executable instruction (or set of instructions) that causes a vehicle to perform an action (e.g., a driving maneuver). The operating commands may include, but are not limited to, instructions for starting the vehicle to move forward, stopping the vehicle to move forward, starting the vehicle to move backward, stopping the vehicle to move backward, accelerating, decelerating, making a left turn, and making a right turn. In an embodiment, the calculation processor 146 is similar to the processor 304 described below with reference to fig. 3. Examples of devices 101 include a steering controller 102, a brake 103, a gear, an accelerator pedal or other acceleration control mechanism, windshield wipers, side door locks, window controls, and steering indicators.

In an embodiment, the AV system 120 includes sensors 121 for measuring or inferring attributes of the state or condition of the AV100, such as the location, linear and angular velocities and accelerations, and heading (e.g., direction of the front end of the AV 100) of the AV. Examples of sensors 121 include GPS, and Inertial Measurement Units (IMU) that measure vehicle linear acceleration and angular rate, wheel rate sensors for measuring or estimating wheel slip rate, wheel brake pressure or torque sensors, engine torque or wheel torque sensors, and steering angle and angular rate sensors.

In an embodiment, the sensors 121 further comprise sensors for sensing or measuring properties of the environment of the AV. Such as a monocular or stereo camera 122 for the visible, infrared, or thermal (or both) spectrum, LiDAR 123, RADAR, ultrasonic sensors, time-of-flight (TOF) depth sensors, rate sensors, temperature sensors, humidity sensors, and precipitation sensors.

In an embodiment, the AV system 120 includes a data storage unit 142 and a memory 144 for storing machine instructions related to the computer processor 146 or data collected by the sensors 121. In an embodiment, the data storage unit 142 is similar to the ROM 308 or the storage device 310 described below in connection with FIG. 3. In an embodiment, memory 144 is similar to main memory 306 described below. In an embodiment, data storage unit 142 and memory 144 store historical, real-time, and/or predictive information about environment 190. In an embodiment, the stored information includes maps, driving performance, traffic congestion updates, or weather conditions. In an embodiment, data related to the environment 190 is transmitted to the AV100 via a communication channel from the remote database 134.

In an embodiment, the AV system 120 includes a communication device 140 for communicating measured or inferred attributes of the state and conditions of other vehicles (such as position, linear and angular velocities, linear and angular accelerations, and linear and angular headings, etc.) to the AV 100. These devices include vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I) communication devices as well as devices for wireless communication over point-to-point or ad hoc (ad hoc) networks or both. In embodiments, communication devices 140 communicate across the electromagnetic spectrum (including radio and optical communications) or other media (e.g., air and acoustic media). The combination of vehicle-to-vehicle (V2V), vehicle-to-infrastructure (V2I) communications (and in some embodiments one or more other types of communications) is sometimes referred to as vehicle-to-everything (V2X) communications. The V2X communications are generally compliant with one or more communication standards for communications with and between autonomous vehicles.

In an embodiment, the communication device 140 comprises a communication interface. For example, these devices may include wired, wireless, WiMAX, Wi-Fi, Bluetooth, satellite, cellular, optical, near-field, infrared, or radio interfaces. The communication interface transmits data from the remote database 134 to the AV system 120. In an embodiment, remote database 134 is embedded in cloud computing environment 200, as described in fig. 2. The communication interface 140 transmits data collected from the sensors 121 or other data related to the operation of the AV100 to the remote database 134. In an embodiment, the communication interface 140 transmits information related to remote operation to the AV 100. In some embodiments, the AV100 communicates with other remote (e.g., "cloud") servers 136.

In an embodiment, the remote database 134 also stores and transmits digital data (e.g., data storing road and street locations, etc.). These data are stored in memory 144 on AV100 or transmitted from remote database 134 to AV100 over a communications channel.

In an embodiment, the remote database 134 stores and transmits historical information (e.g., velocity and acceleration rate profiles) related to driving attributes of vehicles that previously traveled along the trajectory 198 at similar times of the day. In one implementation, such data may be stored in memory 144 on AV100 or transmitted from remote database 134 to AV100 over a communications channel.

A computing device 146 located on the AV100 algorithmically generates control actions based on real-time sensor data and a priori information so that the AV system 120 can perform its autonomous driving capabilities.

In an embodiment, the AV system 120 includes a computer peripheral 132 connected to a computing device 146 for providing information and reminders to and receiving input from a user (e.g., an occupant or remote user) of the AV 100. In an embodiment, peripheral 132 is similar to display 312, input device 314, and cursor controller 316 discussed below with reference to fig. 3. The connection is wireless or wired. Any two or more of the interface devices may be integrated into a single device.

In an embodiment, the AV system 120 receives and enforces a privacy level of the occupant, for example, specified by the occupant or stored in a profile associated with the occupant. The privacy level of the occupant determines how to permit use of particular information associated with the occupant (e.g., occupant comfort data, biometric data, etc.) stored in the occupant profile and/or stored on the cloud server 136 and associated with the occupant profile. In an embodiment, the privacy level specifies particular information associated with the occupant that is deleted once the ride is completed. In an embodiment, the privacy level specifies particular information associated with the occupant and identifies one or more entities authorized to access the information. Examples of the designated entities that are authorized to access the information may include other AVs, third party AV systems, or any entity that may potentially access the information.

The privacy level of the occupant may be specified at one or more levels of granularity. In an embodiment, the privacy level identifies the particular information to be stored or shared. In an embodiment, the privacy level applies to all information associated with the occupant, such that the occupant may specify not to store or share her personal information. The designation of entities permitted to access particular information may also be specified at various levels of granularity. The various entity sets that are permitted to access particular information may include, for example, other AVs, cloud servers 136, particular third party AV systems, and so forth.

In an embodiment, the AV system 120 or the cloud server 136 determines whether the AV100 or another entity has access to certain information associated with the occupant. For example, a third party AV system attempting to access occupant inputs related to a particular spatiotemporal location must obtain authorization, e.g., from the AV system 120 or the cloud server 136, to access information associated with the occupant. For example, the AV system 120 uses the occupant's specified privacy level to determine whether occupant input related to the spatiotemporal location may be presented to a third party AV system, AV100, or another AV. This enables the privacy level of the occupant to specify which other entities are allowed to receive data related to the occupant's actions or other data associated with the occupant.

FIG. 2 illustrates an example "cloud" computing environment. Cloud computing is a service delivery model that allows convenient, on-demand access to a shared pool of configurable computing resources (e.g., networks, network bandwidth, servers, processes, memory, storage, applications, virtual machines, and services) over a network. In a typical cloud computing system, one or more large cloud data centers house machines for delivering services provided by the cloud. Referring now to fig. 2, cloud computing environment 200 includes

cloud data centers

204a, 204b, and 204c interconnected by cloud 202.

Data centers

204a, 204b, and 204c provide cloud computing services for computer systems 206a, 206b, 206c, 206d, 206e, and 206f connected to cloud 202.

Cloud computing environment 200 includes one or more cloud data centers. In general, a cloud data center (e.g., cloud data center 204a shown in fig. 2) refers to a physical arrangement of servers that make up a cloud (e.g., cloud 202 shown in fig. 2 or a particular portion of a cloud). For example, the servers are physically arranged in rooms, groups, rows, and racks in a cloud data center. The cloud data center has one or more zones, including one or more server rooms. Each room has one or more rows of servers, each row including one or more racks. Each rack includes one or more individual server nodes. In some implementations, servers in a zone, room, rack, and/or row are grouped into groups according to physical infrastructure requirements of the data center facility (including electrical, energy, thermal, heat, and/or other requirements). In an embodiment, the server node is similar to the computer system described in FIG. 3. Data center 204a has a number of computing systems distributed across multiple racks.

Cloud 202 includes

cloud data centers

204a, 204b, and 204c and network resources (e.g., network devices, nodes, routers, switches, and network cables) for connecting

cloud data centers

204a, 204b, and 204c and facilitating access to cloud computing services by computing systems 206 a-f. In an embodiment, the network represents any combination of one or more local networks, wide area networks, or internet networks connected by wired or wireless links deployed using terrestrial or satellite connections. Data exchanged over the network is transmitted using a variety of network layer protocols, such as Internet Protocol (IP), multi-protocol label switching (MPLS), Asynchronous Transfer Mode (ATM), frame relay (FrameRelay), etc. Further, in embodiments where the network represents a combination of multiple sub-networks, a different network layer protocol is used on each underlying sub-network. In some embodiments, the network represents one or more interconnected internetworks (e.g., the public internet, etc.).

Computing systems 206a-f or cloud computing service consumers are connected to cloud 202 through network links and network adapters. In embodiments, computing systems 206a-f are implemented as various computing devices, such as servers, desktops, laptops, tablets, smartphones, internet of things (IoT) devices, autonomous vehicles (including cars, drones, space shuttles, trains, buses, and the like), and consumer electronics. In embodiments, computing systems 206a-f are implemented in or as part of other systems.

Fig. 3 illustrates a computer system 300. In an implementation, the computer system 300 is a special purpose computing device. Special purpose computing devices are hardwired to perform the techniques, or include digital electronic devices such as one or more Application Specific Integrated Circuits (ASICs) or Field Programmable Gate Arrays (FPGAs) that are persistently programmed to perform the techniques, or may include one or more general purpose hardware processors programmed to perform the techniques according to program instructions in firmware, memory, other storage, or a combination. Such dedicated computing devices may also incorporate custom hardwired logic, ASICs or FPGAs with custom programming to accomplish these techniques. In various embodiments, the special purpose computing apparatus is a desktop computer system, portable computer system, handheld device, network device, or any other device that contains hard-wired and/or program logic to implement these techniques.

In an embodiment, computer system 300 includes a bus 302 or other communication mechanism for communicating information, and a hardware processor 304 coupled with bus 302 for processing information. The hardware processor 304 is, for example, a general purpose microprocessor. Computer system 300 also includes a main memory 306, such as a Random Access Memory (RAM) or other dynamic storage device, coupled to bus 302 for storing information and instructions to be executed by processor 304. In one implementation, main memory 306 is used for storing temporary variables or other intermediate information during execution of instructions to be executed by processor 304. When stored in a non-transitory storage medium accessible to processor 304, these instructions cause computer system 300 to become a special-purpose machine that is customized to perform the operations specified in the instructions.

In an embodiment, computer system 300 further includes a Read Only Memory (ROM)308 or other static storage device coupled to bus 302 for storing static information and instructions for processor 304. A storage device 310, such as a magnetic disk, optical disk, solid state drive, or three-dimensional cross-point memory, is provided and connected to bus 302 for storing information and instructions.

In an embodiment, computer system 300 is coupled via bus 302 to a display 312, such as a Cathode Ray Tube (CRT), Liquid Crystal Display (LCD), plasma display, Light Emitting Diode (LED) display, or Organic Light Emitting Diode (OLED) display for displaying information to a computer user. An input device 314, including alphanumeric and other keys, is coupled to bus 302 for communicating information and command selections to processor 304. Another type of user input device is cursor control 316, such as a mouse, a trackball, touch display, or cursor direction keys for communicating direction information and command selections to processor 304 and for controlling cursor movement on display 312. Such input devices typically have two degrees of freedom in two axes, a first axis (e.g., the x-axis) and a second axis (e.g., the y-axis), that allow the device to specify positions in a plane.

According to one embodiment, the techniques herein are performed by computer system 300 in response to processor 304 executing one or more sequences of one or more instructions contained in main memory 306. Such instructions are read into main memory 306 from another storage medium, such as storage device 310. Execution of the sequences of instructions contained in main memory 306 causes processor 304 to perform the process steps described herein. In alternative embodiments, hard-wired circuitry may be used in place of or in combination with software instructions.

The term "storage medium" as used herein refers to any non-transitory medium that stores data and/or instructions that cause a machine to function in a particular manner. Such storage media includes non-volatile media and/or volatile media. Non-volatile media includes, for example, optical disks, magnetic disks, solid state drives, or three-dimensional cross-point memories, such as storage device 310. Volatile media includes dynamic memory, such as main memory 306. Common forms of storage media include, for example, a floppy disk, a flexible disk, hard disk, solid state drive, magnetic tape, or any other magnetic data storage medium, a CD-ROM, any other optical data storage medium, any physical medium with a hole pattern, a RAM, a PROM, and EPROM, a FLASH-EPROM, NV-RAM, or any other memory chip or cartridge.

Storage media is distinct from but may be used in combination with transmission media. Transmission media participate in the transfer of information between storage media. For example, transmission media includes coaxial cables, copper wire and fiber optics, including the wires that comprise bus 302. Transmission media can also take the form of acoustic or light waves, such as those generated during radio-wave and infra-red data communications.

In an embodiment, various forms of media are involved in carrying one or more sequence of instructions to processor 304 for execution. For example, the instructions may initially be carried on a magnetic disk or solid state drive of a remote computer. The remote computer loads the instructions into its dynamic memory and sends the instructions over a telephone line using a modem. A modem local to computer system 300 receives the data on the telephone line and uses an infra-red transmitter to convert the data to an infra-red signal. An infra-red detector receives the data carried in the infra-red signal and appropriate circuitry places the data on bus 302. Bus 302 carries the data to main memory 306, from which main memory 306 processor 304 retrieves and executes the instructions. The instructions received by main memory 306 may optionally be stored on storage device 310 either before or after execution by processor 304.

Computer system 300 also includes a communication interface 318 coupled to bus 302. Communication interface 318 provides a multi-way, two-way data communication coupling to a network link 320 that is connected to a local network 322. For example, communication interface 318 is an Integrated Services Digital Network (ISDN) card, cable modem, satellite modem, or a modem to provide a data communication connection to a corresponding type of telephone line. As another example, communication interface 318 is a Local Area Network (LAN) card to provide a data communication connection to a compatible LAN. In some implementations, a wireless link is also implemented. In any such implementation, communication interface 318 sends and receives electrical, electromagnetic or optical signals that carry digital data streams representing various types of information.

Network link 320 typically provides data communication through one or more networks to other data devices. For example, network link 320 provides a connection through local network 322 to a host computer 324 or to a cloud data center or equipment operated by an Internet Service Provider (ISP) 326. ISP 326 in turn provides data communication services through the world wide packet data communication network now commonly referred to as the "internet". Local network 322 and internet 328 both use electrical, electromagnetic or optical signals that carry digital data streams. The signals through the various networks and the signals on network link 320 and through communication interface 318 are exemplary forms of transmission media, with communication interface 318 carrying digital data to and from computer system 300. In an embodiment, network 320 comprises cloud 202 or a portion of cloud 202 as described above.

Computer system 300 sends messages and receives data, including program code, through the network(s), network link 320 and communication interface 318. In an embodiment, computer system 300 receives code for processing. The received code may be executed by processor 304 as it is received, and/or stored in storage device 310, or other non-volatile storage for later execution.

Autonomous vehicle architecture

Fig. 4 illustrates an example architecture 400 for an autonomous vehicle (e.g., AV100 shown in fig. 1). Architecture 400 includes a perception module 402 (sometimes referred to as a perception circuit), a planning module 404 (sometimes referred to as a planning circuit), a control module 406 (sometimes referred to as a control circuit), a positioning module 408 (sometimes referred to as a positioning circuit), and a database module 410 (sometimes referred to as a database circuit). Each module plays a role in the operation of the AV 100. Collectively, the

modules

402, 404, 406, 408, and 410 may be part of the AV system 120 shown in fig. 1. In some embodiments, any of the

modules

402, 404, 406, 408, and 410 are a combination of computer software (e.g., executable code stored on a computer-readable medium) and computer hardware (e.g., one or more microprocessors, microcontrollers, application specific integrated circuits [ ASICs ], hardware memory devices, other types of integrated circuits, other types of computer hardware, or a combination of any or all of these).

Modules

402, 404, 406, 408, and 410 are each sometimes referred to as processing circuitry (e.g., computer hardware, computer software, or a combination of both). Combinations of any or all of the

modules

402, 404, 406, 408, and 410 are also examples of processing circuitry.

In use, the planning module 404 receives data representing the destination 412 and determines data representing a trajectory 414 (sometimes referred to as a route) that the AV100 can travel in order to reach (e.g., arrive at) the destination 412. In order for planning module 404 to determine data representing trajectory 414, planning module 404 receives data from perception module 402, positioning module 408, and database module 410.

The perception module 402 identifies nearby physical objects using, for example, one or more sensors 121 as also shown in fig. 1. The objects are classified (e.g., grouped into types such as pedestrian, bicycle, automobile, traffic sign, etc.), and a scene description including the classified objects 416 is provided to the planning module 404.

The planning module 404 also receives data representing the AV location 418 from the positioning module 408. The positioning module 408 determines the AV location by using data from the sensors 121 and data (e.g., geographic data) from the database module 410 to calculate the location. For example, the positioning module 408 uses data from GNSS (global navigation satellite system) sensors and geographic data to calculate the longitude and latitude of the AV. In an embodiment, the data used by the positioning module 408 includes high precision maps with lane geometry attributes, maps describing road network connection attributes, maps describing lane physics attributes such as traffic rate, traffic volume, number of vehicle and bicycle lanes, lane width, lane traffic direction, or lane marker types and locations, or combinations thereof, and maps describing spatial locations of road features such as crosswalks, traffic signs, or other travel signals of various types, and the like. In an embodiment, a high accuracy map is constructed by adding data to a low accuracy map via automatic or manual annotation.

The control module 406 receives data representing the track 414 and data representing the AV location 418. The control module 406 operates the control functions 420 a-420 c of the AV (e.g., steering, throttle, braking, ignition) in such a way that the AV100 travel track 414 will reach the destination 412. For example, if the trajectory 414 includes a left turn, the control module 406 will operate the control functions 420 a-420 c as follows: the steering angle of the steering function will cause the AV100 to turn left and the throttle and brakes will cause the AV100 to pause and wait for a passing pedestrian or vehicle before making a turn.

Autonomous vehicle input

FIG. 5 illustrates examples of inputs 502a-502d (e.g., sensors 121 shown in FIG. 1) and outputs 504a-504d (e.g., sensor data) used by the perception module 402 (FIG. 4). One input 502a is a LiDAR (light detection and ranging) system (e.g., LiDAR 123 shown in FIG. 1). LiDAR is a technology that uses light (e.g., a line of light such as infrared light) to obtain data related to a physical object in its line of sight. The LiDAR system generates LiDAR data as output 504 a. For example, LiDAR data is a collection of 3D or 2D points (also referred to as point clouds) used to construct a representation of the environment 190.

The other input 502b is a RADAR system. RADAR is a technology that uses radio waves to obtain data about nearby physical objects. RADAR may obtain data related to objects that are not within a line of sight of the LiDAR system. The RADAR system 502b generates RADAR data as output 504 b. For example, RADAR data is one or more radio frequency electromagnetic signals used to construct a representation of the environment 190.

Another input 502c is a camera system. Camera systems use one or more cameras (e.g., digital cameras using light sensors such as charge coupled devices CCD) to acquire information about nearby physical objects. The camera system generates camera data as output 504 c. The camera data is generally in the form of image data (e.g., data in an image data format such as RAW, JPEG, PNG, etc.). In some examples, the camera system has multiple independent cameras, for example for the purpose of stereoscopic imagery (stereo vision), which enables the camera system to perceive depth. Although the object perceived by the camera system is described herein as "nearby," this is with respect to AV. In use, the camera system may be configured to "see" objects that are far away (e.g., as far as 1 kilometer or more in front of the AV). Thus, the camera system may have features such as sensors and lenses optimized for sensing distant objects.

Another input 502d is a Traffic Light Detection (TLD) system. TLD systems use one or more cameras to obtain information about traffic lights, street signs, and other physical objects that provide visual navigation information. The TLD system generates TLD data as output 504 d. The TLD data often takes the form of image data (e.g., data in an image data format such as RAW, JPEG, PNG, etc.). The TLD system differs from the system containing the camera in that: TLD systems use cameras with a wide field of view (e.g., using a wide-angle lens or a fisheye lens) to obtain information about as many physical objects as possible that provide visual navigation information, so that the AV100 has access to all relevant navigation information provided by these objects. For example, the viewing angle of a TLD system may be about 120 degrees or more.

In some embodiments, the outputs 504a-504d are combined using sensor fusion techniques. Thus, the individual outputs 504a-504d may be provided to other systems of the AV100 (e.g., to the planning module 404 as shown in fig. 4), or in the form of a single combined output or multiple combined outputs of the same type (e.g., using the same combining technique or combining the same output or both) or of different types (e.g., using different individual combining techniques or combining different individual outputs or both). In some embodiments, early fusion techniques are used. Early fusion techniques were characterized by: the outputs are combined, after which one or more data processing steps are applied to the combined output. In some embodiments, post-fusion techniques are used. The later stage fusion technology is characterized in that: after applying one or more data processing steps to the individual outputs, the outputs are combined.

FIG. 6 illustrates an example of a LiDAR system 602 (e.g., input 502a shown in FIG. 5). The LiDAR system 602 emits light 604a-604c from a light emitter 606 (e.g., a laser emitter). Light emitted by LiDAR systems is typically not in the visible spectrum; for example, infrared light is often used. Some of the emitted light 604b encounters a physical object 608 (e.g., a vehicle) and is reflected back to the LiDAR system 602. (light emitted from a LiDAR system does not typically penetrate physical objects, e.g., solid form physical objects.) the LiDAR system 602 also has one or more light detectors 610 for detecting reflected light. In an embodiment, one or more data processing systems associated with a LiDAR system generate an image 612 that represents a field of view 614 of the LiDAR system. The image 612 includes information representing the boundary 616 of the physical object 608. Thus, the image 612 is used to determine the boundaries 616 of one or more physical objects in the vicinity of the AV.

FIG. 7 shows the LiDAR system 602 in operation. In the scenario shown in this figure, the AV100 receives camera system output 504c in the form of an image 702 and LiDAR system output 504a in the form of LiDAR data points 704. In use, the data processing system of the AV100 compares the image 702 with the data points 704. In particular, the physical object 706 identified in the image 702 is also identified in the data points 704. In this way, the AV100 perceives the boundaries of the physical object based on the contours and densities of the data points 704.

FIG. 8 shows additional details of the operation of a LiDAR system 602. As described above, the AV100 detects boundaries of physical objects based on characteristics of data points detected by the LiDAR system 602. As shown in FIG. 8, a flat object, such as the ground 802, will reflect the light 804a-804d emitted from the LiDAR system 602 in a consistent manner. In other words, because the LiDAR system 602 emits light using consistent intervals, the ground 802 will reflect light back to the LiDAR system 602 at the same consistent intervals. As the AV100 travels on the ground 802, the LiDAR system 602 will continue to detect light reflected by the next valid waypoint 806 without blocking the road east and west. However, if the object 808 blocks the road, the light 804e-804f emitted by the LiDAR system 602 will reflect from the points 810a-810b in a manner that is not consistent with expectations. From this information, the AV100 can determine that an object 808 exists.

Autonomous vehicle planning

Fig. 9 illustrates a block diagram 900 of the relationship between the inputs and outputs of planning module 404 (e.g., as illustrated in fig. 4). Generally, the output of the planning module 404 is a route 902 from a starting point 904 (e.g., a source location or an initial location) to an ending point 906 (e.g., a destination or a final location). Route 902 is typically defined by one or more road segments. For example, a road segment refers to a distance to be traveled through at least a portion of a street, road, highway, driveway, or other physical area suitable for a car to travel. In some examples, if AV100 is an off-road vehicle such as a four-wheel drive (4WD) or all-wheel drive (AWD) car, SUV, or minitruck, for example, route 902 includes "off-road" road segments such as unpaved roads or open fields.

In addition to the route 902, the planning module outputs lane-level route planning data 908. The lane-level routing data 908 is used to travel through segments of the route 902 at particular times based on the conditions of the segments. For example, if the route 902 includes a multi-lane highway, the lane-level routing data 908 includes trajectory planning data 910, where the AV100 can use the trajectory planning data 910 to select a lane from among multiple lanes, for example, based on whether an exit is adjacent, whether there are other vehicles in more than one of the lanes, or other factors that change over the course of several minutes or less. Likewise, in some implementations, the lane-level routing data 908 includes rate constraints 912 that are specific to a section of the route 902. For example, if the road segment includes pedestrians or unexpected traffic, the rate constraint 912 may limit the AV100 to a slower than expected rate of travel, such as a rate based on the speed limit data for the road segment.

In an embodiment, inputs to planning module 404 include database data 914 (e.g., from database module 410 shown in fig. 4), current location data 916 (e.g., AV location 418 shown in fig. 4), destination data 918 (e.g., for destination 412 shown in fig. 4), and object data 920 (e.g., classified object 416 as perceived by perception module 402 shown in fig. 4). In some embodiments, database data 914 includes rules used in planning. The rules are specified using a formal language (e.g., using boolean logic). In any given situation encountered by the AV100, at least some of these rules will apply to that situation. A rule is applicable to a given situation if the rule has a condition satisfied based on information available to the AV100 (e.g., information related to the surrounding environment). The rules may have priority. For example, the rule of "move to the leftmost lane if the highway is an expressway" may have a lower priority than "move to the rightmost lane if the exit is close within one mile".

Fig. 10 illustrates a directed graph 1000 used in path planning (e.g., by planning module 404 (fig. 4)). In general, a directed graph 1000, such as the directed graph shown in FIG. 10, is used to determine a path between any starting point 1002 and ending point 1004. In the real world, the distance separating the start 1002 and end 1004 may be relatively large (e.g., in two different metropolitan areas), or may be relatively small (e.g., two intersections adjacent a city block or two lanes of a multi-lane road).

In an embodiment, directed graph 1000 has nodes 1006a-1006d representing different places AV100 may occupy between a start point 1002 and an end point 1004. In some examples, nodes 1006a-1006d represent segments of a road, for example, where the start point 1002 and the end point 1004 represent different metropolitan areas. In some examples, for example, where the start point 1002 and the end point 1004 represent different locations on the same road, the nodes 1006a-1006d represent different locations on the road. Thus, the directed graph 1000 includes information at different levels of granularity. In an embodiment, a directed graph with high granularity is also a subgraph of another directed graph with a larger scale. For example, most information of a directed graph with a starting point 1002 and an ending point 1004 that are far away (e.g., many miles away) is at a low granularity, and the directed graph is based on stored data, but the directed graph also includes some high granularity information for a portion of the directed graph that represents a physical location in the field of view of the AV 100.

In an embodiment, at low granularity, objects 1008a-1008b represent areas that the car cannot travel through, such as areas without streets or roads. At high granularity, objects 1008a-1008b represent physical objects in the field of view of AV100, such as other cars, pedestrians, or other entities with which AV100 cannot share a physical space. In embodiments, some or all of the objects 1008a-1008b are static objects (e.g., objects that do not change location, such as street lights or utility poles, etc.) or dynamic objects (e.g., objects that are capable of changing location, such as pedestrians or other cars, etc.).

Nodes 1006a-1006d are connected by edges 1010a-1010 c. If two nodes 1006a-1006b are connected by an edge 1010a, the AV100 may travel between one node 1006a and the other node 1006b, e.g., without having to travel to an intermediate node before reaching the other node 1006 b. (when referring to AV100 traveling between nodes, meaning that AV100 travels between two physical locations represented by respective nodes.) edges 1010a-1010c are generally bi-directional in the sense that AV100 travels from a first node to a second node, or from a second node to a first node. In an embodiment, edges 1010a-1010c are unidirectional in the sense that AV100 may travel from a first node to a second node, whereas AV100 may not travel from the second node to the first node. The edges 1010a-1010c are unidirectional where the edges 1010a-1010c represent individual lanes of, for example, a one-way street, road, or highway, or other feature that can only be driven in one direction due to legal or physical constraints.

In an embodiment, planning module 404 uses directed graph 1000 to identify a path 1012 made up of nodes and edges between start point 1002 and end point 1004.

Edges 1010a-1010c have associated costs 1014a-1014 b. The costs 1014a-1014b are values or tuples of values representing the resources that would be spent if the AV100 selected the edge. A typical resource is time. For example, if one edge 1010a represents twice the physical distance as represented by the other edge 1010b, the associated cost 1014a of the first edge 1010a may be twice the associated cost 1014b of the second edge 1010 b. Other factors that affect time include anticipated traffic, number of intersections, speed limits, etc. Another typical resource is fuel economy. The two edges 1010a-1010b may represent the same physical distance, but one edge 1010a may require more fuel than the other edge 1010b, e.g., due to road conditions, expected weather, etc.

When the planning module 404 identifies a path 1012 between the start point 1002 and the end point 1004, the planning module 404 typically selects a path that is optimized for cost, e.g., a path having a minimum total cost when adding the individual costs of the edges together. In other embodiments, other operations may be applied.

Autonomous vehicle control

Fig. 11 illustrates a block diagram 1100 of inputs and outputs of the control module 406 (e.g., as shown in fig. 4). The control module operates in accordance with a controller 1102, the controller 1102 including, for example: one or more processors (e.g., one or more computer processors such as a microprocessor or microcontroller, or both) similar to processor 304; short-term and/or long-term data storage devices (e.g., memory random access memory or flash memory or both) similar to main memory 306, ROM 308, and storage device 310; and instructions stored in the memory that, when executed (e.g., by one or more processors), perform the operations of the controller 1102.

In an embodiment, the controller 1102 receives data representing a desired output 1104. The desired output 1104 generally includes speed, such as speed and heading. The desired output 1104 may be based on, for example, data received from the planning module 404 (e.g., as shown in fig. 4). Depending on the desired output 1104, the controller 1102 generates data that can be used as a throttle input 1106 and a steering input 1108. The throttle input 1106 represents the magnitude of a throttle (e.g., acceleration control) that engages the AV100 to achieve the desired output 1104, such as by engaging a steering pedal or engaging another throttle control. In some examples, the throttle input 1106 also includes data that can be used to engage a brake (e.g., deceleration control) of the AV 100. Steering input 1108 represents a steering angle, such as an angle at which steering control of the AV (e.g., a steering wheel, a steering angle actuator, or other function for controlling the steering angle) should be positioned to achieve the desired output 1104.

In an embodiment, the controller 1102 receives feedback for use in adjusting the inputs provided to the throttle and steering. For example, if the AV100 encounters a disturbance 1110, such as a hill, the measured rate 1112 of the AV100 drops below the desired output rate. In an embodiment, any measured output 1114 is provided to the controller 1102 such that the required adjustments are made, for example, based on the difference 1113 between the measured rate and the desired output. The measurement outputs 1114 include a measurement location 1116, a measurement speed 1118 (including speed and heading), a measurement acceleration 1120, and other outputs measurable by sensors of the AV 100.

In an embodiment, information related to the disturbance 1110 is detected in advance, for example, by a sensor such as a camera or LiDAR sensor, and provided to the predictive feedback module 1122. The predictive feedback module 1122 then provides information to the controller 1102 that the controller 1102 can use to adjust accordingly. For example, if a sensor of the AV100 detects ("sees") a hill, the controller 1102 may use this information to prepare to engage the throttle at the appropriate time to avoid significant deceleration.

Fig. 12 shows a block diagram 1200 of the inputs, outputs, and components of a controller 1102. The controller 1102 has a rate analyzer 1202 that affects the operation of a throttle/brake controller 1204. For example, the rate analyzer 1202 instructs the throttle/brake controller 1204 to accelerate or decelerate using the throttle/brake 1206 based on feedback received by the controller 1102 and processed by the rate analyzer 1202, for example.

The controller 1102 also has a lateral tracking controller 1208 that affects the operation of the steering wheel controller 1210. For example, the lateral tracking controller 1208 instructs the steering wheel controller 1210 to adjust the position of the steering angle actuator 1212, based on feedback received by the controller 1102 and processed by the lateral tracking controller 1208, for example.

The controller 1102 receives a number of inputs for determining how to control the throttle/brake 1206 and the steering angle actuator 1212. The planning module 404 provides information used by the controller 1102 to, for example, select a heading at which the AV100 is to begin operation and determine which road segment to drive through when the AV100 reaches an intersection. The positioning module 408 provides information describing the current location of the AV100 to the controller 1102, for example, so that the controller 1102 can determine whether the AV100 is in a location that is expected to be based on the manner in which the throttle/brake 1206 and steering angle actuator 1212 are being controlled. In an embodiment, the controller 1102 receives information from other inputs 1214, such as information received from a database, a computer network, or the like.

Modeling human driving behavior through uncontrolled traffic intersections

For autonomous vehicles, driving through uncontrolled traffic intersections is challenging for various reasons. For example, an autonomous vehicle must determine whether there is sufficient time to merge to avoid a collision, in addition to calculating the velocity of the traversing vehicle in real time. These calculations must be done in real time, which requires either a large amount of processing power in the autonomous vehicle, or a fast connection to the server when processing is done at the server. Thus, generating a model (e.g., a set of rules) for driving through an uncontrolled traffic intersection enables the autonomous vehicle to use less processing power and/or network connectivity.

Fig. 13 illustrates an example of an autonomous vehicle attempting to merge at an uncontrolled traffic intersection. The autonomous vehicle 1302 is illustrated waiting to merge into the intersection 1300. At this point, the autonomous vehicle 1302 must make a calculation to determine whether to merge into the intersection. In particular, the autonomous vehicle 1302 must determine whether merging into the intersection 1300 would cause the autonomous vehicle 1302 to collide with one or more of the

other vehicles

1304 and 1306 approaching the intersection 1300. Additionally, the autonomous vehicle 1302 may determine whether merging into the intersection 1300 may result in one or more of the other vehicles (e.g., vehicles 1034 and 1306) slowing down and/or stopping suddenly. The autonomous vehicle 1302 must balance these two determinations with being overly conservative and thus waiting too long at each intersection. The vehicle must make these determinations (e.g., determine whether a human driver will maneuver in a particular set of situations) in view of occupant comfort and maximizing traffic throughput. In one implementation, when making the determination, the autonomous vehicle 1302 uses an autonomous vehicle driving model that models human behavior through the uncontrolled traffic intersection.

Process 1400 of FIG. 14 illustrates actions that a computer system may take in order to generate an autonomous vehicle driving model to drive through an uncontrolled traffic intersection. FIG. 14 and other figures in the present invention are described with reference to a computer system that performs these acts. However, those skilled in the art will appreciate that the acts of fig. 14 and other figures may be performed by disclosed components (e.g., components local to the autonomous vehicle such as the processor 146, the perception module 402, the planning module 404, the control module 406, or another suitable component). Alternatively, some modules of the computer system may be located remotely from the autonomous vehicle, and some actions may be performed remotely and sent to the autonomous vehicle, while other actions may be performed at the autonomous vehicle.

In block 1405, the computer system determines driving behaviors of a plurality of manually-operated vehicles (e.g., vehicles driven by a human operator and whose driving behaviors are affected by the driving skills of the human operator), each of which has engaged in a traffic merge behavior at a respective uncontrolled traffic intersection (e.g., a portion of a first traffic lane where traffic from a second lane is merged without direction of a traffic controller such as a traffic light or traffic police). Driving behavior represents a driving reaction of a human operator intending to merge into an uncontrolled traffic intersection or observing the merging of another vehicle into an uncontrolled traffic intersection. Fig. 13 illustrates one example of an uncontrolled traffic intersection where a vehicle 1302 is attempting to merge at the uncontrolled traffic intersection.

In an embodiment, the computer system, in determining the autonomous vehicle driving model, is configured to determine whether the manually operated vehicle is attempting to merge with traffic at the intersection or is not attempting to merge with traffic at the intersection. For example, the computer system may iterate through the data sets of traffic consolidation behaviors and determine whether the manually operated vehicles attempt consolidation or do not attempt consolidation. The computer system may separate data associated with attempted merges from data associated with non-attempted merges. As discussed below, the computer system may use one or both of these data sets to generate a model.

In an embodiment, when determining driving behavior, the computer system identifies a vehicle incorporating a corresponding uncontrolled traffic intersection (e.g., a "go" decision), and determines: (1) the respective velocities of the plurality of other vehicles being driven toward the respective uncontrolled traffic intersection, (2) the distance of each of the plurality of other vehicles from the merged vehicle, and (3) the lanes of each of the other vehicles. For example, a computer system may receive data collected from one or more uncontrolled traffic intersections and generate a data set corresponding to a vehicle that is merging (e.g., a "forward decision") based on the collected data. In an embodiment, the model may be trained at the intersection level. That is, the model is saved for a particular intersection, and when a vehicle approaches the particular intersection, the intersection is detected and the model is retrieved and used for the intersection. In an embodiment, the model may be trained at the city level. That is, the model is saved for a particular city and when an autonomous vehicle is detected in the city, the model is retrieved and used within the city. In embodiments, the model may be used at the legal domain level or the national level. That is, the model is taken out and used when the autonomous vehicle is within the scope of the law or in a specific country. Those skilled in the art will appreciate that the usage or level of usage of the model may depend on the driving laws and/or the homogeneity of the driving behaviour within the area.

In block 1410, the computer system determines, based on the driving behavior, an autonomous vehicle driving model that describes autonomous vehicle driving behavior (e.g., merging or not merging) to be achieved by autonomous vehicle driving to merge with traffic at the uncontrolled traffic intersection. For example, for a given speed and heading of the autonomous vehicle, the model indicates that the autonomous vehicle is either heading (i.e., merging into a traffic intersection) or not heading (i.e., waiting).

In an embodiment, a computer system analyzes data associated with driving behavior and identifies different scenarios in which manually operated vehicles merge into an intersection. That is, the computer system may analyze the distance of each other vehicle approaching the intersection from the intersection and the velocity of each vehicle. Based on this information, the computer system can build a model (e.g., one or more rules) for merging into the intersection. The rule may indicate that the autonomous vehicle should continue to merge into the uncontrolled traffic intersection when each other vehicle traveling toward the intersection is at least a distance away and traveling at a rate that is not faster than a certain rate. In an embodiment, the computer system generates one or more rules based on the velocity and distance of other vehicles approaching the intersection.

In an embodiment, the computer system may consider data related to manually operated vehicles that are not incorporated into an intersection. Specifically, the computer system identifies vehicles that do not merge into a corresponding uncontrolled traffic intersection (e.g., a "no go" decision). For each of the vehicles that are not incorporated into the respective uncontrolled traffic intersection, a respective velocity of other vehicles in the vicinity of the each vehicle that are driving toward the respective uncontrolled traffic intersection, a distance of each of the other vehicles from the each vehicle that is not being merged, and a lane of each of the other vehicles are determined. For example, the computer system may filter data for instances where the manually operated vehicle is not incorporated into an intersection. The computer system may analyze the speed and distance of each of the other vehicles traveling toward the intersection. Based on the analysis, the computer system may generate a model (or add one or more rules to a model generated in view of data for manually operated vehicles that are incorporated into the intersection). For example, a rule may represent: an autonomous vehicle should not merge into an intersection if one or more vehicles are traveling toward the intersection at a rate greater than a certain rate and closer than a certain distance from the intersection. In an embodiment, a computer system generates a plurality of rules (e.g., a rule for velocity and a rule for distance).

In an embodiment, the computer system takes into account the lane in which each approaching vehicle is located when generating or adding to the model. For example, the computer system may generate a rule that represents a combination of the speed of the approaching vehicle, the distance of the approaching vehicle from the intersection, and the lane of the approaching vehicle (e.g., whether the approaching vehicle is located in the lane that the manually operated vehicle is attempting to merge).

In an embodiment, the computer system generates a model based on both data related to manually operated vehicles that incorporate an uncontrolled traffic intersection and that do not incorporate an uncontrolled traffic intersection, taking into account the rates of other vehicles driving toward the intersection, the distances of these other vehicles from the intersection, and the lanes associated with each of these other vehicles. For example, the computer system may calculate a maximum velocity and a maximum distance for each other vehicle when the manually operated vehicle merges into the intersection and base the rules of the model on the maximum velocity and/or the maximum distance. The rule may indicate that the autonomous vehicle should merge into the intersection if the distance of each vehicle from the intersection is less than the maximum distance of an approaching vehicle from the intersection and/or the velocity of each approaching vehicle is less than the maximum velocity. In an embodiment, the rules of the model may include the following rules based on the collected data: the speed and distance rules may be overridden if the approaching vehicle is not in the lane that the autonomous vehicle is attempting to merge.

In an embodiment, the computer system uses the collected data relating to the manually operated vehicle as input to the neural network to train the neural network for when to merge into an uncontrolled intersection. The inputs may include velocity data, distance data, and lane data for each approaching vehicle.

In an embodiment, the autonomous vehicle driving model maps (map) the velocity of other vehicles driving toward an intersection, the distance of each other vehicle from the particular vehicle, and the lanes of each other vehicle for the particular vehicle driving toward an uncontrolled traffic intersection. For example, the computer system may iterate through the data and store the speed, distance, and lane of the approaching vehicle in a separate data structure. The data mapped to the map may be added to the model so that it may be processed while the autonomous vehicle is attempting to merge into an uncontrolled intersection. In an embodiment, the data may be plotted to prepare the data to be input into the neural network as training data.

In one embodiment, a tracking system (e.g., a sensor system that includes a plurality of sensors and senses the driving of a vehicle) is located at an uncontrolled traffic intersection (or multiple tracking systems of multiple traffic intersections). The computer system receives sensed results of tracking driving behavior from the tracking systems. The computer system may update the autonomous vehicle driving model based on the received data or as the data is received.

In an embodiment, a computer system generates a baseline model for an autonomous vehicle driving model. In particular, the computer system collects data (e.g., received data associated with a merge behavior of a manually operated vehicle), identifies a dataset from the collected data, and fits the dataset to the baseline model. In an embodiment, the computer system uses one of the likelihood ratios (Cox and Snell, Nagelkerke, McFadden, or Tjur) to generate the baseline model.

As the process 1400 of fig. 14 continues, in block 1415, the computer system uses the control circuitry to operate the autonomous vehicle according to the autonomous vehicle driving model. In an embodiment, a computer system is located in an autonomous vehicle and generates driving instructions according to an autonomous vehicle driving model. In particular, the computer system determines a distance from the autonomous vehicle to another vehicle driving toward the uncontrolled traffic intersection and a velocity of the other vehicle (e.g., based on the sensor input). The computer system processes the distance and the velocity according to the autonomous vehicle driving model, and provides instructions to the control circuit to merge or not merge traffic at the uncontrolled traffic intersection based on results of processing the distance and the velocity. For example, the autonomous vehicle may take speed, distance as input, and output a determination regarding whether the autonomous vehicle should continue merging into the intersection or should not merge based on this information. If the output indicates that the autonomous vehicle should merge, the computer system generates one or more driving commands for the autonomous vehicle to merge into the intersection. If the output indicates that the autonomous vehicle should not merge into the intersection, the computer system begins the determination process again when the other vehicle passes through the intersection.

In an embodiment, a non-transitory computer-readable storage medium is provided. The non-transitory computer readable storage medium stores one or more programs for execution by one or more processors of an apparatus, the one or more programs including instructions, which when executed by the one or more processors, cause the apparatus to perform the actions described above.

In embodiments, the computer system is separate from the autonomous vehicle and may be located remotely from the autonomous vehicle or on the autonomous vehicle. The computer system is configured to determine driving behavior of a plurality of manually operated vehicles, each manually operated vehicle having engaged in traffic merging behavior at a respective uncontrolled traffic intersection. The autonomous vehicle includes one or more computer-readable media storing computer-executable instructions and one or more processors communicatively coupled to a computer system and configured to execute the computer-executable instructions.

In an embodiment, an autonomous vehicle receives driving behavior from a computer system and determines an autonomous vehicle driving model based on the driving behavior, the autonomous vehicle driving model describing autonomous vehicle driving behavior to be achieved by the autonomous vehicle for traffic merging behavior at an uncontrolled traffic intersection. The autonomous vehicle may determine the autonomous vehicle driving model based on any of the methods described above. The autonomous vehicle then operates itself according to the autonomous vehicle driving model. In an embodiment, the computer system is separate from the vehicle and may provide the driving behavior to the vehicle. In an embodiment, the computer system is onboard a vehicle.

In an embodiment, a computer system predicts what operations a human operator will do in a particular set of situations. For example, if an autonomous vehicle is driving and it is detected that another vehicle (e.g., a human operated vehicle) is attempting to merge into an intersection, the computer system may use the same model to determine whether a human operator will attempt to merge or wait. The computer system may drive the vehicle according to the determination. Thus, if the computer system determines, based on the model, that the vehicle is to continue merging, the computer system may slow the vehicle slightly to avoid the collision. In an embodiment, the computer system accelerates the autonomous vehicle if the computer system determines that another vehicle is predicted to wait until the autonomous vehicle passes through the intersection. In an embodiment, the computer system may generate different models with different rules, but make the determination based on the same data.

Modeling human driving behavior to drive beyond stationary vehicles

For autonomous vehicles, attempting to drive beyond stationary vehicles in the same lane is challenging for a variety of reasons. For example, the autonomous vehicle, in addition to determining its own velocity and distance to the stationary vehicle, must also consider any vehicles in the adjacent lane(s) to determine if there is sufficient distance and/or time to pass the stationary vehicle. These calculations must be done in real time, which requires either significant processing power in the autonomous vehicle, or a quick connection to the server when processing is done at the server. Thus, generating a model (e.g., a set of rules) for driving beyond a stationary vehicle enables the autonomous vehicle to use less processing power and/or network connectivity.

Fig. 15 shows an example of an autonomous vehicle attempting to pass a stationary vehicle in the same lane. The autonomous vehicle 1502 is illustrated as determining that conditions are appropriate for passing a stationary vehicle in the same lane. Before overriding a maneuver, the autonomous vehicle 1502 must make calculations to determine where it is best or safe to pass the stationary vehicle 1504. In particular, the autonomous vehicle 1502 must determine whether an overrun by a vehicle in an adjacent lane would cause the autonomous vehicle 1502 to hit another vehicle that may be in an adjacent lane, and whether there is sufficient longitudinal and lateral distance available to make the overrun. In an embodiment, autonomous vehicle 1502 must balance this determination with stopping too conservatively and multiple times behind a stopped vehicle. The vehicle must make these determinations (e.g., determine whether a human driver will maneuver in a particular set of situations) in view of occupant comfort and maximum driving time. In one implementation, in making the determination, the autonomous vehicle 1502 uses an autonomous vehicle driving model that models human behavior when attempting to override a stationary vehicle in the same lane.

Process 1600 of fig. 16 shows actions that a computer system may take in order to generate an autonomous vehicle driving model to drive beyond a stationary vehicle in the same lane. FIG. 16 and other figures in the present invention are described with reference to a computer system that performs these acts. However, those skilled in the art will appreciate that the acts of fig. 16 and other figures may be performed by disclosed components (e.g., components local to the autonomous vehicle such as the processor 146, the perception module 402, the planning module 404, the control module 406, or another suitable component). Alternatively, some modules of the computer system may be located remotely from the autonomous vehicle, and some actions may be performed remotely and sent to the autonomous vehicle, while other actions may be performed at the autonomous vehicle.

In block 1605, the computer system determines driving behaviors of a plurality of manually operated vehicles (e.g., vehicles that are being driven by a human operator and whose driving behaviors are affected by the driving skills of the human operator), each of which has been driven past a stationary vehicle in the same lane (e.g., a stationary vehicle may be parked or paused in the same lane as a manually operated vehicle, and traffic in adjacent lanes may be heading in the same or opposite direction as the direction the autonomous vehicle is heading for). Driving behavior represents the driving reaction of a human operator intending to drive a vehicle that is beyond (i) in front of the vehicle of the human operator, (ii) in the same lane as the vehicle of the human operator, and (iii) stationary.

In an embodiment, a computer system receives data describing driving behavior of a manually operated vehicle. These data may include one or more of a speed of the manually operated vehicles, a heading of the manually operated vehicles, a longitudinal distance between each manually operated vehicle and the corresponding stationary vehicle, a lateral distance when each manually operated vehicle is driving past a stationary vehicle, a direction of an adjacent lane, and other suitable data. In an embodiment, the model may be trained at the road level. That is, the model is saved for a particular road, and when an autonomous vehicle is on the particular road, the road is detected and the model is retrieved and used for the road. In an embodiment, the model may be trained at the city level. That is, the model is saved for a particular city and when an autonomous vehicle is detected in that city, the model is retrieved and used within that city. In embodiments, the model may be used at the legal domain level or the national level. That is, the model is taken out and used when the autonomous vehicle is within a particular jurisdiction or in a particular country. Those skilled in the art will appreciate that the usage or level of usage of the model may depend on the driving laws and/or the homogeneity of the driving behaviour within the area.

In an embodiment, determining driving behavior comprises: it is determined that a lane change has been made for each manually operated vehicle to drive beyond the corresponding stationary vehicle. For example, the computer system may analyze the received data associated with the manually operated vehicle and determine that the manually operated vehicle changed lanes to drive beyond a subset of the data of the stationary vehicle (e.g., rather than stopping behind the stationary vehicle). In an embodiment, data associated with scenarios where a manually operated vehicle does not change lanes to drive beyond a stationary vehicle is filtered out, as this data may not be useful in certain scenarios. However, in various embodiments, data relating to vehicles that do not change lanes (e.g., stop behind a stationary vehicle) is useful for determining where (e.g., at what rate and distance) an autonomous vehicle should stop rather than attempting to drive past a stationary vehicle.

In an embodiment, determining driving behavior comprises: when each manually operated vehicle initiates a lane change to an adjacent lane, a longitudinal distance between each manually operated vehicle and the corresponding stationary vehicle is determined. For example, fig. 15 shows one example of the longitudinal distance. Determining the longitudinal distance of each manually operated vehicle enables the system to assess at what distance the human driver has initiated a lane change and to mimic this behavior via an autonomous driving vehicle model. For example, the computer system may analyze the longitudinal distance of each manually operated vehicle and generate a model (e.g., one or more rules) for changing lanes based on the distance from the stationary vehicle. The computer system may calculate one or more of mean, median, and mode distances based on the distance, and base one or more rules on the distances. For example, a rule may represent: the autonomous vehicle should initiate a lane change at a particular distance from the stationary vehicle, where the particular distance may be an average, mode, or median distance calculated from data associated with driving behavior.

In an embodiment, determining driving behavior comprises: the rate at which each manually operated vehicle initiates a lane change is determined. For example, the computer system may analyze the rates corresponding to each manually-operated vehicle and determine one or more of the mean, mode, and median rates at which the manually-operated vehicle initiated a lane change to drive past the stationary vehicle. The computer system may generate rules based on the rate or add rules to the model based on the rate. For example, a rule may represent: the control circuitry should slow the autonomous vehicle to a lower rate if the speed of the autonomous vehicle is greater than one of the mode, average, or median speed.

In an embodiment, determining driving behavior comprises: as each manual vehicle drives past the stationary vehicle, a lateral distance between each stationary vehicle and the respective manual vehicle in the adjacent lane is determined. For example, the computer system may analyze how close each manually operated vehicle is to the respective stationary vehicle as it passes by the stationary vehicle. The computer system may determine one or more of a mean, median, and mode lateral distance for each instance of the manually operated vehicle passing the respective stationary vehicle. The computer system may generate and/or update an autonomous vehicle driving model based on the information. That is, the computer system may create a rule indicating that the autonomous vehicle should maneuver in a manner that enables the autonomous vehicle to traverse the stationary vehicle at a lateral distance that is no less than one of the determined distances.

In an embodiment, determining driving behavior comprises: the direction of traffic in lanes adjacent to the lanes of each manually operated vehicle is determined. The data relating to the direction of traffic in adjacent lanes may represent: whether a human driver takes into account the direction of traffic when performing a particular maneuver. For example, if the changing characteristics of each particular maneuver (e.g., speed, starting distance, whether to maneuver or stop the vehicle, and/or other suitable characteristics) are within a threshold when the maneuver is performed in places where there are adjacent lanes of traffic in the same direction and in other places where there are adjacent lanes of traffic in different directions, then the computer system determines that the lane direction does not affect the maneuver. However, if the change characteristic reaches or is above the threshold, the computer system determines that the lane direction affects the maneuver. For example, the computer system may analyze each lane-change maneuver of the manually-operated vehicles while driving beyond the stationary vehicle and determine whether each manually-operated vehicle changed lanes into the same direction of lane or a different direction of lane. For example, if the computer system determines that in many or most instances, a manually operated vehicle has changed lanes into lanes of the same direction, the computer system generates or updates a model to change lanes to lanes of the same direction. However, if the adjacent lane is not in the same direction as the lane of the manual vehicle, the computer system may analyze what actions were taken by the driver who manually operated the vehicle and add those actions to the model (e.g., the driver stops and waits).

To continue with process 1600 of fig. 16, in block 1610, the computer system determines an autonomous vehicle driving model that describes autonomous vehicle driving behavior to be achieved by an autonomous vehicle attempting to drive past a stationary vehicle in the same lane based on the driving behavior. For example, the computer system may analyze instances of the manually-operated vehicle and determine a minimum distance (longitudinal distance) between the manually-operated vehicle and the stationary vehicle when the manually-operated vehicle begins the override maneuver. The computer system may generate a portion of the model (e.g., rules) based on the information. The rule may represent: the autonomous vehicle may begin an override maneuver if the autonomous vehicle is at least a minimum longitudinal distance away from the stationary vehicle.

In another example, the computer system may analyze instances of the manually operated vehicle and determine a minimum distance (lateral distance) between the manually operated vehicle and the stationary vehicle as the manually operated vehicle is passing the stationary vehicle. Based on the distance, the computer system may generate a rule (e.g., to include in the autonomous vehicle driving model). For example, the rule may represent: the autonomous vehicle must continue to change to another lane until the minimum lateral distance is reached. When the minimum lateral distance is to be reached (based on the degree of turning), the autonomous vehicle should adjust its heading to complete the lane change.

In an embodiment, the autonomous vehicle driving model is configured to determine when the autonomous vehicle should initiate a lane change to pass a stationary vehicle based on driving behavior. For example, based on driving behavior, the computer system may generate a model (e.g., including one or more rules) for passing stationary vehicles in the same lane. As discussed above, the computer system may determine from the driving behavior data one or more of the mean, mode, median, maximum, and minimum longitudinal distances at which the manually-operated vehicle initiated an override maneuver to traverse the stationary vehicle. The computer system may generate one or more rules based on the information. For example, a rule may represent: the autonomous vehicle may begin an override maneuver when the autonomous vehicle is at least a longitudinal distance away from the stationary vehicle. The longitudinal distance may be an average, a mode, a median, a maximum, a minimum, or another suitable longitudinal distance.

In an embodiment, the computer system takes into account the velocity of each manually operated vehicle when generating the autonomous vehicle driving model. For example, the computer system may determine one or more of an average, a mode (e.g., when rounding the rate), a median, a maximum, and a minimum based on the driving behavior. Based on one or more of these values, the computer system may generate a rule indicating that the autonomous vehicle should accelerate/decelerate to achieve the determined rate before passing by the stationary vehicle. In an embodiment, the computer system generates the rule based on a combination of the velocity and the longitudinal distance. For example, the computer system may analyze the driving behavior of the manually operated vehicles and determine, for each manually operated vehicle, the rate at which the manually operated vehicle is moving and the longitudinal distance at which to begin an override maneuver. Thus, from the pair of velocity and longitudinal distance, the computer system may generate a rule that represents: if the autonomous vehicle is moving at a certain rate, an override maneuver should be initiated at a certain distance (or before reaching a certain distance).

Additionally or alternatively, the computer system may generate a rule (for the autonomous vehicle driving model) that represents: if the autonomous vehicle is too close to the stationary vehicle such that more distance is needed to perform the override maneuver based on the velocity of the autonomous vehicle, the autonomous vehicle should slow down such that the override maneuver may occur within the parameters of the driving behavior.

In an embodiment, the computer system generates/updates the autonomous vehicle driving model in view of the direction of the lane (e.g., adjacent lane) to which the manually operated vehicle has changed lane. For example, the computer system may generate a rule for the model that represents: if the adjacent lane is in a different direction than the lane of the autonomous vehicle, the autonomous vehicle should change lanes (begin to overtake the maneuver) at a minimum longitudinal distance (in some cases, at an increased rate so as to spend as little time as possible in the adjacent lane) and then should change back to the lane in time as the autonomous vehicle passes by the stationary vehicle. In another example, if the adjacent lane is in the same direction as the lane in which the autonomous vehicle is driving, the computer system may generate a rule that represents: the autonomous vehicle should change lanes some time/distance before reaching the minimum distance.

To continue with process 1600 of fig. 16, in block 1615, the computer system uses the control circuit to operate the autonomous vehicle according to the autonomous vehicle driving model. As discussed above, the computer system may be located at the autonomous vehicle or at a location remote from the autonomous vehicle. In the case where the computer system is located at an autonomous vehicle, the computer system may direct other vehicle systems to drive according to an autonomous vehicle driving model. For example, where the computer system is located at a location remote from the autonomous vehicle, the computer system may send commands to the vehicle (e.g., via a wireless network).

In an embodiment, a computer system predicts what operations a human operator will do in a particular set of situations. For example, if an autonomous vehicle must stop in a lane (e.g., due to an occupant needing to exit the vehicle), the system may use the same model to determine whether a vehicle driving in the rear (e.g., taking into account the speed and distance of the vehicle relative to the stopped vehicle) has sufficient distance to drive over or stop. In an embodiment, the computer system may generate different models via different rules, but make the determination based on the same data.

In an embodiment, the computer system operates the autonomous vehicle according to an autonomous vehicle driving model by performing the following actions. The computer system determines a velocity of the autonomous vehicle, a heading of the autonomous vehicle, and a direction of traffic in an adjacent lane. For example, the computer system may receive this information from one or more sensors (e.g., cameras, radrs, LiDAR, and/or other suitable sensors) located in the vehicle (as described above). Using the autonomous vehicle driving model, the computer system determines a longitudinal distance and a lateral distance for passing a stationary vehicle in the same lane. For example, the computer system may access an autonomous vehicle driving model and retrieve one or more rules associated with the model. The rules may include thresholds for one or more of longitudinal distance, lateral distance, and lane direction. The computer system may apply the sensor data to the rule and, based on the application, provide an indication to the autonomous vehicle to initiate a lane change at least when a distance between the autonomous vehicle and the stationary vehicle is greater than a longitudinal distance and traverse the stationary vehicle at least a lateral distance.

In an embodiment, the autonomous vehicle applies the model to a situation where the autonomous vehicle is attempting to pass a moving vehicle ahead, where in that situation adjacent lanes have traffic moving in opposite directions (e.g., on a two lane road containing two lanes with traffic in opposite directions). In this case, the autonomous vehicle must determine whether to continue with the override maneuver or to wait for a better opportunity (e.g., while driving behind a forward vehicle). In this and other embodiments, the autonomous vehicle uses data representing human operator decisions to determine whether to continue to merge adjacent lanes with traffic flowing in different directions. The determination is based on, for example, the velocity of the vehicle ahead, the velocity and/or distance of any oncoming vehicles in adjacent lanes, and traffic regulations (e.g., whether override is allowed, speed limit, and/or another appropriate traffic regulation).

In an embodiment, the computer system receives data from a tracking system located at or near the stationary vehicle (e.g., a sensor system for sensing and sending driving behavior of the manually operated vehicle to the computer system, which may or may not be in the same location as the tracking system) to track the driving behavior. For example, a stationary vehicle may include sensors that track each vehicle attempting to drive beyond, and record the velocity of the other vehicles, the longitudinal distance at which the vehicle begins to exceed a maneuver, the lateral distance at which the maneuver is performed, and other suitable data.

In an embodiment, the computer system for collecting and processing driving behavior data may be separate from the vehicle. For example, the computer system may be a module that can be added to a vehicle, or the computer system may be located remotely from the vehicle. The computer system is configured to determine driving behavior (e.g., the driving behavior represents a driving reaction of a human operator intending to drive a vehicle that is beyond (i) in front of the vehicle of the human operator, (ii) in the same lane as the vehicle of the human operator, and (iii) is stationary) of a plurality of manually operated vehicles (e.g., the manually operated vehicles are vehicles that are being driven by the human operator and whose driving behavior is influenced by the driving skill of the human operator), each manually operated vehicle having been driven beyond a stationary vehicle in the same lane (e.g., the stationary vehicle may be parked or have been parked in the same lane as the manually operated vehicle; traffic in adjacent lanes may proceed in the same or different direction as AV).

In this case, the autonomous vehicle includes one or more computer-readable media (e.g., storage devices including one or more hard disk drives, flash drives) storing computer-executable instructions and one or more processors communicatively coupled to the computer system and configured to execute the computer-executable instructions; the performing includes the following operations. The autonomous vehicle receives driving behavior with the vehicle from the computer system. For example, the autonomous vehicle may receive processed data based on the collected driving behavior. The autonomous vehicle may receive longitudinal distance, velocity, lateral distance, lane direction, and other suitable information.

The autonomous vehicle determines/generates an autonomous vehicle driving model (e.g., for a given speed and heading of the autonomous vehicle, the model indicates that the autonomous vehicle is driving beyond a stationary vehicle) based on the tracked driving behavior, the autonomous vehicle driving model describing an autonomous vehicle driving behavior (e.g., speed and heading of the autonomous vehicle) to be achieved by the autonomous vehicle driving to drive beyond a stationary vehicle in the same lane that the autonomous vehicle is driving. In this case, the autonomous vehicle may use the actions described above to generate a model (e.g., generate rules that use the received data). The autonomous vehicle then operates itself according to the autonomous vehicle driving model.

As discussed above, the computer system may be located at the autonomous vehicle, or may be separate from the autonomous vehicle (e.g., located remotely from the autonomous vehicle). The computer system is separate from the vehicle. In these cases, the computer system may analyze the collected data and provide driving behavior to the autonomous vehicle, which may generate an autonomous vehicle driving model accordingly. In an embodiment, the autonomous vehicle generates an autonomous vehicle driving model taking into account the following maneuvers/driving restrictions/characteristics of the particular autonomous vehicle. For example, the amount of time required for the vehicle to decelerate when the brakes are applied, the weight of the autonomous vehicle, the sensitivity of the sensors (e.g., at what distance a stationary vehicle may be detected), and other suitable limitations/characteristics of the autonomous vehicle.

Modeling human driving behavior to avoid crossing roads

For autonomous vehicles, it is challenging to attempt to avoid crossroad people in the road (whether in the same lane or in one of the lanes near the autonomous vehicle) for various reasons. For example, the autonomous vehicle, in addition to determining its speed and location, must also determine whether there is sufficient distance and/or time to avoid the person crossing the road(s) before or after the person crossing the road(s) is in the lane of the autonomous vehicle (in view of the speed of the autonomous vehicle), or whether the autonomous vehicle should change into an adjacent lane (if an adjacent lane is available). The autonomous vehicle must also take into account any vehicles in the adjacent lane(s). These calculations must be done in real time, which requires either a large amount of processing power in the autonomous vehicle, or must be quickly connected to the server when processing is done at the server. In addition, the autonomous vehicle must determine whether merging into the intersection 1300 will cause the vehicle in the adjacent lane to slow down and/or stop suddenly. The autonomous vehicle must balance these determinations with being overly conservative and waiting too long for crossroad people to leave the lane. The vehicle must make these determinations (e.g., determine whether a human driver will maneuver in a particular set of situations) in view of occupant comfort and maximizing traffic throughput. Thus, generating a model (e.g., a set of rules) for driving in the presence of crossroad people enables the autonomous vehicle to use less processing power and/or network connectivity.

Fig. 17-19 illustrate examples of various scenarios relating to avoiding people crossing roads. Fig. 17 shows a scenario for avoiding a person crossing a road when an autonomous vehicle is driving on a road having two lanes, where adjacent lanes have traffic in the same direction. The data relating to the direction of traffic in adjacent lanes may indicate whether a human driver is considering the direction of traffic when performing a particular maneuver. For example, if the changing characteristics of each particular maneuver (e.g., speed, starting distance, whether to maneuver or stop the vehicle, and/or another suitable characteristic) are within a threshold when the maneuver is performed in places where there are adjacent lanes of traffic in the same direction and in other places where there are adjacent lanes of traffic in different directions, then the computer system determines that the lane direction does not affect the maneuver. However, if the change characteristic reaches or is above the threshold, the computer system determines that the lane direction affects the maneuver. Specifically, fig. 17 shows a road having two lanes (lane 1702 and lane 1704) that allow traffic in the same direction. The autonomous vehicle 1706 is traveling in a lane 1704. A crossroad person 1708 is also present in the same lane (lane 1704) as the vehicle. In this scenario, the autonomous vehicle must determine how to avoid crossroad people. For example, the autonomous vehicle must determine whether to change lanes, slow down, stop, do nothing, perform a different maneuver, or perform a combination of maneuvers. In making the determination, the autonomous vehicle may have to consider factors such as its speed, traffic in adjacent lanes, speed (both speed and direction) of crossroad people, and other suitable factors.

Fig. 18 shows a scenario for avoiding a person crossing a road when an autonomous vehicle is driving on a road having two lanes, where adjacent lanes have traffic in opposite directions. Specifically, fig. 18 shows a road having two lanes (lane 1802 and lane 1804) that allow traffic in different directions. Autonomous vehicle 1806 is traveling in lane 1804. A crossroad person 1808 is also present in the same lane (lane 1804) as the vehicle. In this scenario, the autonomous vehicle must determine how to avoid crossroad people (1808). For example, the autonomous vehicle must determine whether to change lanes (which may be illegal at the location), slow down, stop, do nothing, perform a different maneuver, or perform a combination of maneuvers. In making the determination, the autonomous vehicle may have to consider factors such as its speed, traffic in adjacent lanes, speed (both speed and direction) of crossroad people, traffic regulations, and other suitable factors. In an embodiment, the model may be trained at the road level. That is, the model is saved for a particular road, and when the autonomous vehicle approaches the particular road, the road is detected and the model is retrieved and used for the road. In an embodiment, the model may be trained at the city level. That is, the model is saved for a particular city, and when an autonomous vehicle is detected in that city, the model is retrieved and used within that city. In embodiments, the model may be used at the legal domain level or the national level. That is, the model is taken out and used when the autonomous vehicle is within a particular jurisdiction or in a particular country. Those skilled in the art will appreciate that the usage or level of usage of the model may depend on the driving laws and/or the homogeneity of the driving behaviour within the area.

Fig. 19 illustrates a scenario for avoiding a person crossing a road when an autonomous vehicle is driving on a road with a single lane. Specifically, fig. 19 shows a road having one lane (lane 1902) in which an autonomous vehicle 1906 is traveling. A crossroad person 1908 is also present in the same lane (lane 1902) as the vehicle. In this scenario, the autonomous vehicle must determine how to avoid crossroad people (1908). For example, the autonomous vehicle must determine whether to park on edge (in the presence of a shoulder), slow down, stop, do nothing, perform a different maneuver, or perform a combination of maneuvers. In making the determination, the autonomous vehicle may have to consider factors such as its own speed, the presence of road shoulders, the speed (both speed and direction) of people crossing the road, traffic regulations, and other suitable factors. As referred to herein, the term "shoulder" refers to the ground area on the curb where cars may rest if they are needed.

Process 2000 of fig. 20 illustrates actions that a computer system may take in order to generate an autonomous vehicle driving model to avoid an entity (e.g., a crossroad human, an animal, a rider, or another suitable entity). FIG. 20 and other figures in the present invention are described with reference to a computer system performing these acts. However, those skilled in the art will appreciate that the acts of fig. 20 and other figures may be performed by disclosed components (e.g., components local to the autonomous vehicle such as the processor 146, the perception module 402, the planning module 404, the control module 406, or another suitable component). Alternatively, some modules of the computer system may be located remotely from the autonomous vehicle, and some actions may be performed remotely and sent to the autonomous vehicle, while other actions may be performed at the autonomous vehicle.

In block 2005, the computer system determines driving behaviors of a plurality of manually-operated vehicles (e.g., vehicles that are driven by a human operator and whose driving behaviors are affected by the driving skills of the human operator), each of which has been driven in a lane into which the respective entity entered. For example, driving behavior represents the driving reaction of a human operator observing that a crossroad person enters the driving lane of the human operator's vehicle. A person crossing a road may be a human, an animal, a cyclist or another suitable entity.

In an embodiment, determining driving behavior comprises: the number of lanes constituting the lane in which each manually operated vehicle has been driven is determined. For example, the computer system may determine, for each manually operated vehicle, the number of lanes in the roadway on which the manually operated vehicle is driving. Fig. 17-19 show examples of roads with one lane and two lanes, however, those skilled in the art will appreciate that the present invention contemplates other scenarios with more lanes. This information is important because it can inform the vehicle via the autonomous vehicle driving model which maneuvers can be attempted and which maneuvers cannot be attempted (e.g., change to an adjacent lane if available).

In an embodiment, the computer system determines a direction of traffic in lanes of a road on which each manually operated vehicle is driven. For example, fig. 17 shows two lanes with traffic in the same direction, potentially enabling the vehicle to change to an adjacent lane. Fig. 18 shows two lanes with traffic in different directions, potentially removing lane-change maneuvers from the options an autonomous vehicle may have.

In an embodiment, the computer system, in determining the driving behavior, determines a number of lanes traversed by the respective entity before entering a lane in which each manually operated vehicle is driven. For example, if an entity traverses one or two lanes in a particular direction, the entity may have provided a clue to the human driver that the entity will not return into those lanes. This scenario may result in the operator manually operating the vehicle merging into an adjacent lane that the entity has traversed. The computer system may generate/update an autonomous vehicle driving model based on the information. For example, the computer system may add, based on the information, a rule to the model that represents: if an entity crosses a lane before entering the lane in which the autonomous vehicle is driving, the autonomous vehicle should merge into the lane that the entity has crossed.

In an embodiment, the computer system, when determining the driving behavior, determines an angle at which the respective entity traversed the plurality of lanes before entering the lane in which each manually operated vehicle was driven. For example, the information related to the angle of traversal may have informed the human driver of the trajectory of the entity, thereby enabling avoidance of the entity with maneuvers that avoid the trajectory of the entity. Based on this information, the computer system may generate rules for the model. The rule may entail computing a trajectory of the entity based on the angle (and the velocity of the entity). The rule may represent: based on the trajectory of the entity, the autonomous vehicle should change lanes as much as possible into lanes located in opposite directions of the trajectory of the entity.

In an embodiment, the computer system, in determining the driving behavior, determines that the respective entity traversed the lane at the specified traversal site (e.g., the specified traversal site is a crosswalk or zebra crossing or construction zone). For example, if the traversal location of the entity is a designed traversal location, the computer system may generate a rule for the model to stop. The rules may take into account traffic rules for the location (e.g., allowing a person to cross at a designated crossing location at any time).

In an embodiment, the computer system, in determining the driving behavior, determines that the respective entity is following a trajectory that includes the specified crossing location and a portion of the lane in which each manually operated vehicle is driven, and excludes the specified crossing location. For example, if a pedestrian is crossing at a designated crossing location, the pedestrian may not be a crossroad person, and thus the scenario does not apply. Thus, the computer system may remove this data from the model building process. In an embodiment, the computer system extension specifies an area occupied by the traversal site. That is, the computer system may generate a buffer for data relating to a person determined to be a pedestrian other than a crossroad. For example, when the area is a construction zone, a buffer zone is provided for people crossing the road to walk around the construction zone.

In an embodiment, the computer system determines at least one of a distance between the respective entity and each manually operated vehicle, a velocity of the respective entity, and a heading of the respective entity when determining the driving behavior. For example, the computer system may analyze the velocity of the manually operated vehicle, the velocity and direction of the entity, and determine which maneuver(s) the human driver performed in order to avoid the entity. For example, based on driving behavior, the computer system may generate a rule that represents: if the autonomous vehicle is traveling at a certain rate and is located at a certain distance from the entity, the autonomous vehicle should change lanes into adjacent lanes or slow down (e.g., slow enough for a crossroad person to leave the lane), stop, or take a combination of maneuvers.

To continue with fig. 20, in block 2010, the computer system determines, based on the driving behavior, an autonomous vehicle driving model that describes an autonomous vehicle driving behavior (e.g., speed and heading of the autonomous vehicle) to be achieved by the autonomous vehicle attempting to drive to avoid the entity. For example, the model may include rules that represent maneuvers to be performed by the autonomous vehicle to avoid the entity for a given speed and heading of the autonomous vehicle relative to the entity.

In an embodiment, the computer system, when determining the autonomous vehicle driving model, determines, for each determined speed and heading of each manually operated vehicle, a minimum distance between each manually operated vehicle and the respective entity without modifying vehicle behavior to allow the respective entity to traverse the lane. For example, the model may include the following rules: if the autonomous vehicle is at least some distance away (based on the driving behavior) and the rate of movement (based on the driving behavior) is not faster than a certain rate, then the autonomous vehicle should not take any action (e.g., because the vehicle is driving at the location where the entity is currently present, the entity will be outside the lane of the vehicle).

In an embodiment, the computer system, when generating the autonomous vehicle driving model, is configured to determine when the autonomous vehicle should initiate a lane change to avoid the entity based on the autonomous vehicle driving behavior. For example, the computer system may generate a rule that represents: if an adjacent lane is available when the autonomous vehicle is level with the entity and the entity is not expected to be in the adjacent lane based on the trajectory of the entity, the autonomous vehicle should change lanes to the adjacent lane.

To continue with process 2000 of fig. 20, in block 2015, the computer system uses the control circuit to operate the autonomous vehicle according to an autonomous vehicle driving model. As discussed above, the computer system may be located at the autonomous vehicle or at a location remote from the autonomous vehicle. In the case where the computer system is located at an autonomous vehicle, the computer system may direct other vehicle systems to drive according to an autonomous vehicle driving model. For example, where the computer system is located at a location remote from the autonomous vehicle, the computer system may send commands to the vehicle (e.g., via a wireless network).

In an embodiment, the computer system operates the autonomous vehicle according to the autonomous vehicle driving model by taking the following actions. The computer system determines a speed and heading of the autonomous vehicle and a speed and heading of the entity. Using the autonomous vehicle driving model, the computer system determines a minimum distance between the autonomous vehicle and the entity for the speed and heading of the autonomous vehicle and the speed and heading of the entity. The computer system provides instructions to the autonomous vehicle to modify the driving behavior of the autonomous vehicle (e.g., slow down or initiate a change to a lane that the entity has traversed to avoid the entity).

In an embodiment, the computer system and the vehicle may be part of an autonomous driving system. The computer system may be a separate system (e.g., module) that may be built into the vehicle or may be located at a location remote from the vehicle. In both cases, the computer system is configured to track driving behavior (e.g., the driving behavior represents a driving reaction of a human operator who intends to avoid a pedestrian in the same lane) of a plurality of manually operated vehicles (e.g., a manually operated vehicle is a vehicle that is driven by a human operator and whose driving behavior is influenced by the driving skills of the human operator). In particular, each manually operated vehicle has been driven in a lane into which the respective entity (e.g. the entity may be a pedestrian or an animal) enters.

A vehicle (e.g., configured on-board) includes one or more computer-readable media storing computer-executable instructions and one or more processors communicatively coupled to a computer system and configured to execute the computer-executable instructions, wherein the execution performs the following. The method includes receiving, with the vehicle, driving behavior from the computer system, and determining, with the computer system, an autonomous vehicle driving model (e.g., whether to change lanes or stop for a given speed and heading of the autonomous vehicle) based on the driving behavior. As described above, the autonomous vehicle driving model describes autonomous vehicle driving behaviors (e.g., speed and heading of the autonomous vehicle) to be achieved by the autonomous vehicle driving to avoid the entity. The autonomous vehicle may then operate on its own in accordance with the autonomous vehicle driving model.

In one example, the computer system resides at a remote location (e.g., in a data center removed from the vehicle) and collects and processes driving behavior data collected from manually operated vehicles. The driving behavior data may include various values describing how each manually operated vehicle avoids an entity (e.g., a pedestrian). The computer system sends the driving behavior to the autonomous vehicle, which in turn determines/generates an autonomous vehicle driving model for avoiding the entity. Then, when a physical avoidance situation occurs on the road, the vehicle operates itself according to the model.

In an embodiment, the system includes a tracking system (e.g., a sensor system for sensing and transmitting driving behavior of the vehicle to a computer system). The tracking system may include one or more of components mounted in the manually operated vehicle to track driving behavior and components mounted near the roadway to detect driving behavior.

Using a model of human driving behavior in an autonomous vehicle

Various scenarios for building/generating an autonomous vehicle driving model are detailed herein. However, proper use of these models is critical to the safety of autonomous vehicles. 21-23 illustrate actions performed by a processing circuit (e.g., of an autonomous vehicle) when using a model in the described driving scenario.

Process 2100 of FIG. 21 illustrates actions that an autonomous vehicle may take in order to use an autonomous vehicle driving model to drive through an uncontrolled traffic intersection. Fig. 21 and other figures in the present invention are described with reference to processing circuitry that performs these actions. However, those skilled in the art will appreciate that the acts of fig. 21 and other figures may be performed by other components (e.g., the processor 146, the perception module 402, the planning module 404, the control module 406, or another suitable component). In block 2105, processing circuitry (e.g., processor 146) receives sensor data associated with an environment of an autonomous vehicle. For example, the processing circuitry may receive LiDAR data, RADAR data, camera data, and other suitable data. In block 2110, the processing circuitry (e.g., processor 146) determines that the autonomous vehicle has arrived at the uncontrolled traffic intersection based on the sensor data. For example, the processing circuitry may analyze the camera data in conjunction with the LiDAR data to determine that the autonomous vehicle is at an uncontrolled intersection (e.g., no traffic lights are present).

In block 2115, the processing circuitry retrieves an autonomous vehicle driving model that models driving behavior of a vehicle that is behaving in a traffic merge with traffic at the uncontrolled traffic intersection. The autonomous vehicle driving model is based on data associated with driving behaviors of a plurality of manually operated vehicles, wherein each manually operated vehicle has performed a traffic merging behavior at a respective uncontrolled traffic intersection. For example, the processing circuitry may retrieve the driving model from the memory 144 (fig. 1). In an embodiment, the processing circuitry may retrieve the autonomous vehicle driving model from the database 410. In an embodiment, the model may be trained at the intersection level. That is, the model is saved for a particular intersection and when an autonomous vehicle approaches the particular intersection, the intersection is detected and the model is retrieved and used for the intersection. In an embodiment, the model may be trained at the city level. That is, the model is saved for a particular city, and when an autonomous vehicle is detected in that city, the model is retrieved and used within that city. In embodiments, the model may be used at the legal domain level or the national level. That is, the model is taken out and used when the autonomous vehicle is within a particular jurisdiction or in a particular country. Those skilled in the art will appreciate that the usage or level of usage of the model may depend on the driving laws and/or the homogeneity of the driving behaviour within the area.

In block 2120, the processing circuitry applies the autonomous vehicle driving model to the sensor data. The autonomous vehicle driving model may include one or more rules (e.g., based on driving behavior) and accept input of the vehicle's environment. Instructions within the autonomous driving model may instruct the processing circuit to apply rules to the inputs to output instructions for the autonomous vehicle. Specifically, in block 2125, the processing circuit determines whether to merge with traffic at the uncontrolled traffic intersection based on applying the sensor data to the autonomous vehicle driving model. For example, the processing circuitry may receive an output that continues to merge or waits (e.g., waits for another merge opportunity) as a result of applying the rules of the model to the sensor data. In block 2130, the processing circuit uses the control circuit to operate the autonomous vehicle according to an autonomous vehicle driving model. For example, if the autonomous vehicle driving model outputs instructions to continue, the processing circuit instructs the control circuit to continue merging with the intersection. However, if the autonomous vehicle driving model outputs instructions to wait, the processing circuit instructs the control circuit to wait (e.g., avoid merging into an intersection).

In an embodiment, the processing circuit receives, from one or more sensors, data representing a distance from an autonomous vehicle to an approaching vehicle driving toward an uncontrolled traffic intersection, and data representing a velocity of the approaching vehicle. For example, autonomous vehicles may be equipped with RADAR/LiDAR or a combination of these instruments. These instruments may detect any vehicles approaching the autonomous vehicle and determine the velocity and distance to the autonomous vehicle. In embodiments, these instruments may be used to determine the distance of each other vehicle moving toward the intersection from where the autonomous vehicle at the intersection will merge. For example, the instruments may determine a distance between the autonomous vehicle and a point at the intersection where the autonomous vehicle is to merge, and based on the distance and the distances to each of the other vehicles, the processing circuitry may determine a distance to the point at the intersection.

In an embodiment, the processing circuitry, when applying the autonomous vehicle driving model to the sensor, performs the following actions. The processing circuit retrieves rules from the autonomous vehicle driving model that include distance thresholds and velocity thresholds for incorporation into the uncontrolled traffic intersection. For example, the autonomous vehicle driving model may be stored as a data structure that includes rules (e.g., instantiation objects). Each rule accepts input and includes a condition for the input. The rules output different values based on whether the condition is satisfied. For example, one rule may include conditions for a distance threshold and a velocity threshold. In another example, the conditions may be divided into two rules. The threshold distance in a rule is a value that is added to the model when the model is created. The value is based on the driving behaviour processed by the computer system.

The processing circuitry compares the distance threshold to the distances received from the one or more sensors and compares the velocity threshold to the velocities received from the one or more sensors, and determines, based on the comparison, whether the distances received from the one or more sensors satisfy the distance threshold and whether the velocities received from the one or more sensors match the velocity threshold. From this determination, the processing circuit determines whether one or more other vehicles driving toward the intersection are far enough and slow enough that the autonomous vehicle has time to merge into the intersection. Based on the determination, the rule outputs a result with instructions regarding whether to merge with the traffic.

In an embodiment, the processing circuitry provides commands to the autonomous vehicle to merge or wait for merge with traffic at the uncontrolled traffic intersection based on results of processing the distance and the velocity.

Process 2200 of fig. 22 illustrates actions that an autonomous vehicle may take in order to use an autonomous vehicle driving model to drive beyond a stationary vehicle in the same lane. Fig. 22 and other figures in the present invention are described with reference to processing circuitry that performs these actions. However, those skilled in the art will appreciate that the acts of fig. 22 and other figures may be performed by other components (e.g., the processor 146, the perception module 402, the planning module 404, the control module 406, or another suitable component). In block 2205, processing circuitry (e.g., processor 146) receives sensor data associated with an environment of an autonomous vehicle. For example, the processing circuitry may receive LiDAR data, RADAR data, camera data, and other suitable data. In block 2210, the processing circuitry (e.g., processor 146) determines, based on the sensor data, that the stationary vehicle is located in the same lane as the autonomous vehicle is driving. For example, the processing circuitry may analyze the camera data in conjunction with the LiDAR data and the RADAR data to determine that the stationary vehicle is in the same lane as the autonomous vehicle. Based at least on the LiDAR data and the RADAR data, the processing circuitry may determine a distance to the vehicle and determine that the vehicle is stationary (i.e., not moving).

In block 2215, the processing circuit retrieves an autonomous vehicle driving model that models driving behavior of driving a vehicle that is beyond a stationary vehicle. The autonomous vehicle driving model is based on data associated with driving behavior of a plurality of manually operated vehicles, wherein each manually operated vehicle is driven beyond a respective stationary vehicle in the same lane. For example, the processing circuitry may retrieve the driving model from the memory 144 (fig. 1). In an embodiment, the processing circuitry may retrieve the autonomous vehicle driving model from the database 410.

In block 2220, the processing circuitry applies the autonomous vehicle driving model to the sensor data. The autonomous vehicle driving model may include one or more rules (e.g., based on driving behavior) and accept input of the vehicle's environment. Instructions within the autonomous driving model may instruct the processing circuit to apply rules to the inputs to output instructions for the autonomous vehicle. Specifically, in block 2225, the processing circuitry determines one or more driving instructions for driving beyond a stationary vehicle in the same lane in which the autonomous vehicle is driving based on applying the sensor data to the autonomous vehicle driving model. For example, the processing circuitry may receive an output of one or more instructions as a result of applying rules of the model to the sensor data. For example, the instructions may include one or more of decelerating and changing lanes to adjacent lanes. Under certain conditions, the processing circuitry instructs the autonomous vehicle to stop. In block 2230, the processing circuit uses the control circuit to operate the autonomous vehicle according to the autonomous vehicle driving model. For example, if the autonomous vehicle driving model outputs an instruction to change lanes, the processing circuit instructs the control circuit to make the lane change.

In an embodiment, the processing circuitry receives one or more of a distance of the stationary vehicle, a velocity of the autonomous vehicle, a distance between the stationary vehicle and another vehicle located in an adjacent lane of the stationary vehicle, a direction of traffic in the adjacent lane, and other suitable data from the one or more sensors. For example, the processing circuit may indicate to each available sensor what data the processing circuit needs for operation according to the model. When the processing circuitry receives the data, the processing circuitry applies the data to the autonomous vehicle driving model.

In an embodiment, the processing circuitry performs the following actions when applying the autonomous vehicle driving model to the sensor data. The processing circuit inputs the received sensor data into an autonomous vehicle driving model. The data includes one or more of a distance to the stationary vehicle, a velocity of the autonomous vehicle, a distance between the stationary vehicle and another vehicle located in an adjacent lane of the stationary vehicle, and a direction of traffic in the adjacent lane. For example, the processing circuitry may load the autonomous driving model into random access memory and execute a process that executes instructions associated with the autonomous driving model. The instructions may include routines for accepting input of the data indicated above. When input is received from the autonomous vehicle driving model, the routine of the model executes the model (e.g., one or more rules) on the data. For example, the autonomous vehicle driving model may be stored as a data structure that includes rules (e.g., instantiation objects). Each rule accepts input and includes a condition for the input. The rules output different values based on whether the condition is satisfied. For example, one rule may include conditions for a distance threshold and a velocity threshold. In another example, the conditions may be divided into two rules. The threshold distance in a rule is a value that is added to the model when the model is created. This value is based on the driving behavior that the computer system handles as described above. The processing circuit receives an output from the autonomous vehicle driving model that includes one or more commands for driving beyond the stationary vehicle. For example, when entering velocity and distance into the rules, the output may be one or more instructions (e.g., change lane to adjacent lane, accelerate, decelerate, stop, or another suitable instruction).

Process 2300 of fig. 23 illustrates actions that an autonomous vehicle may take in order to use an autonomous vehicle driving model to avoid entities (e.g., pedestrians) in the same lane. Fig. 23 and other figures in the present invention are described with reference to processing circuitry that performs these actions. However, those skilled in the art will appreciate that the acts of fig. 23 and other figures may be performed by other components (e.g., the processor 146, the perception module 402, the planning module 404, the control module 406, or another suitable component). In block 2305, the processing circuit (e.g., processor 146) receives sensor data associated with an environment of the autonomous vehicle. For example, the processing circuitry may receive LiDAR data, RADAR data, camera data, and other suitable data. At block 2310, the processing circuit (e.g., the processor 146) determines, based on the sensor data, that the entity has entered a lane in which the autonomous vehicle is driving. For example, the processing circuitry may analyze the camera data in conjunction with the LiDAR data and the RADAR data to determine that the pedestrian is in the same lane as the autonomous vehicle. Based at least on the LiDAR data and RADAR data, the processing circuitry may determine a distance to the pedestrian, a speed (both speed and direction) of the pedestrian, and other suitable information.

In block 2315, the processing circuit retrieves an autonomous vehicle driving model that models driving behavior to avoid entities in the same lane (i.e., in the trajectory of the vehicle). The autonomous vehicle driving model is based on data associated with driving behaviors of a plurality of manually operated vehicles, each of which is driven to avoid an entity (e.g., a pedestrian, an animal, or another suitable entity) in the same lane. For example, the processing circuitry may retrieve the driving model from the memory 144 (fig. 1). In an embodiment, the processing circuitry may retrieve the autonomous vehicle driving model from the database 410.

In block 2320, the processing circuitry applies the autonomous vehicle driving model to the sensor data. The autonomous vehicle driving model may include one or more rules (e.g., based on driving behavior) and accept input of the vehicle's environment (e.g., vehicle's velocity, distance to the entity, speed of the entity (both velocity and direction), and other suitable information). Instructions within the autonomous driving model may instruct the processing circuit to apply rules to the inputs to output instructions for the autonomous vehicle. Specifically, in block 2325, the processing circuitry determines one or more driving instructions to avoid entities in the same lane in which the autonomous vehicle is driving based on applying the sensor data to the autonomous vehicle driving model. For example, the processing circuitry may receive an output of one or more instructions as a result of applying rules of the model to the sensor data. For example, the instructions may include one or more of decelerating and changing lanes to adjacent lanes. Under certain conditions, the processing circuitry indicates that the autonomous vehicle is parked. In block 2330, the processing circuit uses the control circuit to operate the autonomous vehicle according to the autonomous vehicle driving model. For example, if the autonomous vehicle driving model outputs an instruction to change lanes, the processing circuit instructs the control circuit to make the lane change.

In an embodiment, the sensor data associated with the environment of the autonomous vehicle comprises: one or more of a velocity of the autonomous vehicle, a distance from the autonomous vehicle to the entity, a number of available lanes of the autonomous vehicle, a direction of traffic in each of the available lanes, a number of lanes the entity traversed before entering the lane in which the autonomous vehicle is driving, whether the entity is designating a point of traversal across lanes, a speed of the entity, an angle at which the entity traverses the plurality of lanes before entering the lane in which the autonomous vehicle is driving, and other suitable sensor data.

In an embodiment, the processing circuitry performs the following actions when applying the autonomous vehicle driving model to the sensor data. The processing circuitry inputs into the autonomous vehicle driving model one or more of: a speed of the autonomous vehicle, a distance from the autonomous vehicle to the entity, a number of available lanes of the autonomous vehicle, a direction of traffic in each of the available lanes, a number of lanes the entity traverses before entering the lane in which the autonomous vehicle is driving, whether the entity is designating a point of traversal across the lanes, a speed of the entity, and an angle at which the entity traverses multiple lanes before entering the lane in which the autonomous vehicle is driving. The processing circuitry receives an output from the autonomous vehicle driving model that includes one or more commands for avoiding the entity.

For example, an autonomous vehicle may be driving in a center lane of a three lane highway, where all three lanes have traffic in the same direction. If the processing circuit detects a pedestrian on the same (intermediate) lane, an autonomous vehicle driving model may be used to obtain instructions for avoiding the pedestrian. The processing circuitry receives data relating to the speed and direction of the pedestrian (e.g., moving from left to right), as well as the speed and distance to the pedestrian of the autonomous vehicle. The processing circuitry inputs the data into the model. The model may have rules (e.g., generated based on the driving behavior of a manually controlled vehicle) that represent: if the autonomous vehicle is at least a distance away from the pedestrian and the autonomous vehicle's rate of travel is slower than a certain rate while the pedestrian is moving from left to right, the autonomous vehicle should change lanes to the left lane to avoid the pedestrian. The model outputs the instructions and the processing circuitry sends the instructions to the control circuitry of the vehicle for execution.

In an embodiment, the autonomous vehicle driving model is generated by processing circuitry configured on the vehicle. Processing circuitry configured on the autonomous vehicle collects driving behavior of a plurality of manually operated vehicles. Data collection may include receiving data from a central source or the manually operated vehicle itself. In an embodiment, the autonomous vehicle uses its sensors to collect data as it drives on the same road as the manual vehicle. The collected data may include data for one or more scenes. For example, the collected data may include driving behavior of a plurality of manually operated vehicles performing traffic consolidation behavior at respective uncontrolled intersections. The data may also include driving behavior of the plurality of manually operated vehicles driving past a stationary vehicle in the same lane and/or driving behavior of the plurality of manually operated vehicles having been driven to avoid an entity in the same lane.

When the data is received, the processing circuitry generates a corresponding autonomous vehicle driving model. The control circuitry may update the model based on the received supplemental data. As discussed above, the autonomous vehicle driving model may be stored as a set of instructions and a set of data. The data and instructions may be generated on the autonomous vehicle itself or at a location remote from the autonomous vehicle.

Generating models of human driving behavior using neural networks

As previously described herein, human driving behavior may be used as a model of autonomous vehicle driving behavior for certain situations (e.g., incorporating an uncontrolled intersection, driving past stationary vehicles in the same lane, and driving around pedestrians in the same lane). However, human driving behavior may be used as a model for other situations that an autonomous vehicle may encounter on a road. Neural networks may be configured and trained for many situations in which human driving behavior data is available. Neural networks can be used as models for autonomous vehicles in many situations on the road.

Process 2400 of FIG. 24 illustrates actions that a computer system may take in order to generate a model of human driving behavior for a particular traffic maneuver. In block 2405, the computer system receives a data set describing driving behavior of a plurality of manually operated vehicles, each manually operated vehicle attempting to perform a vehicle maneuver, wherein the data set includes a plurality of entries and a plurality of decisions made by a human driver in performing the respective vehicle maneuver, and wherein each entry includes a plurality of fields for a plurality of factors associated with the respective vehicle maneuver. For example, a field may be a tabular column storing data values for aspects of a maneuver such as a velocity of a vehicle, a distance to an object, and another suitable value. The factors may be specific to a maneuver, such that when a human decides to merge or not merge into an intersection, the factors may be one or more distances of one or more other cars to the intersection. In models that model avoidance of pedestrians, the distance to the pedestrian may be a factor.

FIG. 25 shows one possible data set describing the driving behavior of a plurality of manually operated vehicles. The data set 2502 includes a plurality of columns (e.g., fields) having corresponding data values. For example, column 2504 represents the speed of one of the vehicles approaching the intersection, and column 2506 represents whether the human driver is merging or avoiding merging (e.g., go/no-go decision) in view of the data. In addition, fig. 25 shows a scene that incorporates an uncontrolled intersection. Vehicle 2508 is a vehicle that attempts to merge into an uncontrolled intersection.

Vehicles

2510, 2512, and 2514 are other vehicles approaching uncontrolled intersections. The speed of these vehicles and the distance of these vehicles are represented by data in data set 2502.

In block 2410, the computer system fetches a plurality of fields from the dataset. For example, the columns shown in FIG. 25 (e.g.,

columns

2504, 2506, and other columns) may be fields that are fetched by the computer system. In an embodiment, a computer system parses a data set to retrieve fields. In another embodiment, the computer system makes an API call to fetch a field. The data set may be stored in a memory (e.g., random access memory) or a data storage unit (e.g., data storage unit 142 of fig. 1 or database 410 of fig. 4). In an embodiment, a processing circuit (e.g., a processor) retrieves a received data set from a data storage unit and copies the data set into a memory for access and processing.

In an embodiment, a computer system may take the following actions when fetching multiple fields from a dataset. The computer system executes a command requesting a plurality of fields using an Application Programming Interface (API) associated with the data set and receives a plurality of identifiers associated with the plurality of fields in response to the command. For example, a data set may be accessed by a set of APIs, where the APIs may be used to extract various information from the data set. The computer system may use these APIs to extract multiple identifiers.

In block 2415, the computer system identifies a plurality of data types associated with the plurality of fields. For example, as shown in FIG. 25, a data set may include different fields. Some of these fields may be of numeric type (i.e., include numeric values). Some fields may be of Boolean type (i.e., include Boolean data). Other fields may be of a type that includes a limited set. For example, a field may have one of four values (e.g., a string of characters, an alphanumeric character, and other suitable values). A computer system may identify a data type by accessing a data set and determining which kind of value is in each column (i.e., field). In an embodiment, a computer system may generate and execute an API call to obtain the type of each field.

In an embodiment, a computer system identifies multiple data types associated with multiple fields by performing the following actions. The computer system selects, for each of a plurality of fields, an identifier associated with the respective field. For example, each field may include an identifier. The identifier may be a field name (e.g., vehicle speed). In various embodiments, the identifier may be an alphanumeric string, hexadecimal digits, or other suitable identifier. The computer system compares the selected identifiers to a plurality of identifiers associated with known data types. For example, the computer system may access a storage location where the identifier is stored (e.g., in a table) along with the corresponding type. The computer system may iterate through the identifiers and compare the identifiers to stored identifiers, and when a match is found, the computer system identifies a respective data type for each of the plurality of fields based on the comparison.

In block 2420, the computer system determines a plurality of characteristics associated with the traffic maneuver based on the plurality of fields. In an embodiment, each field is characteristic of a traffic maneuver. For example, fig. 25 illustrates characteristics such as the type of turn, the vehicles entering, the distance to the vehicles, the velocity or speed of the vehicles, and other characteristics. FIG. 25 also shows the results of the maneuver (e.g., whether a human driver is merging into the intersection).

In block 2425, the computer system configures the neural network using the plurality of characteristics and the plurality of data types. For example, a computer system may configure the input, hidden, and output layers of a neural network. The hidden layer may have several layers depending on the output requirements for a particular maneuver. For example, if the yes/no decision is to expect an output (e.g., whether to continue through an uncontrolled intersection), only one hidden layer may be needed to process the input. However, if more complex outputs are required (e.g., driving instructions to bypass a stationary vehicle in the same lane or a pedestrian in the same lane), the computer system may configure more than one hidden layer. In addition, the computer system also configures an input layer to the neural network. The computer system may configure the input layer with a plurality of input parameters and types of input parameters. For example, when configuring the input layer of a neural network, a computer system may use multiple characteristics and multiple data types retrieved from a data set. The computer system may use characteristics such as the velocity of the autonomous vehicle, the distance to a particular entity (e.g., a stationary vehicle or a pedestrian), the velocity of the entity, the direction of travel of the entity, and other suitable characteristics. The computer system may also configure the neural network using the type of characteristic.

In an embodiment, a computer system configures a number of inputs based on a number of characteristics of a plurality of characteristics associated with a traffic maneuver, and selects a plurality of initiation weights for each characteristic of the plurality of characteristics. For example, the weight of each distance from each other vehicle to the intersection may be a ratio of 0 to 1 of.6 for the decision regarding whether to merge into the intersection. These weights are changed based on the training set when training the neural network. The computer system then generates a neural network, for example, using an Application Program Interface (API) function. The API function takes as input a plurality of characteristics, a type of field associated with each of the plurality of characteristics, and a type of neural network.

In block 2430, the computer system inputs data associated with the data set into the neural network according to each of the plurality of characteristics. When creating a neural network, the neural network is trained using the data sets. Thus, the data of the data set is input into the neural network such that each characteristic of the neural network corresponds to a field in the data set. For example, when data is input into the neural network, characteristics of vehicle speed in the neural network match fields of vehicle speed in the data set.

In block 2435, the computer system executes a training routine for the neural network using the input data and a plurality of characteristics associated with the traffic maneuver. For example, the computer system may perform a function corresponding to a training routine. The function may be performed using the data set as an input. In an embodiment, a function may enable inputting a path to a data set from which the function is able to read training data.

In an embodiment, the neural network is trained at a location remote from any autonomous vehicle (e.g., at a central location, a factory, or another suitable location). Training the neural network outside the autonomous vehicle may have various advantages. For example, in scenarios where the processing power of an autonomous vehicle is limited, the neural network can be trained faster. In addition, training the neural network outside of the vehicle (e.g., at a data center) enables fast distribution of the trained network (i.e., each vehicle is not required to train each neural network). Thus, in embodiments where the neural network is trained external to the autonomous vehicle, the neural network is sent to one or more autonomous vehicles as needed.

In embodiments, in generating and training neural networks, autonomous vehicles may use these neural networks in driving scenarios designed for these autonomous vehicles. In particular, a computer system (e.g., at an autonomous vehicle) receives sensor data from one or more sensors of the autonomous vehicle. In an embodiment, the computer system may be located on an autonomous vehicle. For example, the computer system may be a module configured on an autonomous vehicle and may be connected with other components of the autonomous vehicle (e.g.,

components

142, 144, 146 and other components associated with fig. 1). The computer system may have some or all of the components of fig. 3. In an embodiment, a computer system may be connected with the components of FIG. 4. In an embodiment, the model may be trained at the intersection/road level. That is, the model is held for a specific intersection/road, and when a vehicle approaches the specific intersection/road, the intersection/road is detected and the model is taken out and used for the intersection/road. In an embodiment, the model may be trained at the city level. That is, the model is saved for a particular city and when an autonomous vehicle is detected in that city, the model is retrieved and used within that city. In embodiments, the model may be used at the legal domain level or the national level. That is, the model is taken out and used when the autonomous vehicle is within a particular jurisdiction or in a particular country. Those skilled in the art will appreciate that the usage or level of usage of the model may depend on the driving laws and/or the homogeneity of the driving behaviour within the area.

In embodiments, the computer system may be located external to the autonomous vehicle and may send information between the vehicle and the computer system carrying the model via the wireless network. For example, an autonomous vehicle may collect sensor data while it is driving and may send the data to a central location. At the central location, the computer system may process the data and send driving instructions to the autonomous vehicle.

In an embodiment, the computer system selects a traffic maneuver to navigate through a portion of the trajectory based on the sensor data. For example, the computer system may receive sensor data (e.g., remotely on an autonomous vehicle) and determine that a particular maneuver needs to be performed to continue traveling on a path to a desired destination. The computer system may determine whether there is an available neural network associated with the selected maneuver. The computer system extracts one or more values corresponding to a plurality of characteristics from the sensor data and executes a neural network using one or more of the plurality of characteristics as input data. For example, if the desired maneuver is being incorporated into the uncontrolled intersection, the computer system extracts from the sensor data characteristics such as the velocity of the autonomous vehicle, the velocity and direction of movement of each other vehicle approaching the uncontrolled intersection, the distance of each other vehicle to the uncontrolled intersection, the lane of each other vehicle, and the direction of each lane. In embodiments, other characteristics may be extracted from the sensor data. The neural network is then executed using the input data.

The computer system receives an output from the neural network and operates the autonomous vehicle according to the output of the neural network using the control circuit. For example, in a scenario where it is necessary to decide whether to proceed to an uncontrolled intersection or wait for other vehicles or entities to pass, the computer system may output a boolean value indicating whether the autonomous vehicle is to proceed or wait. In an embodiment, the output may be one or more commands to continue advancing. For example, the command may include the rate at which the autonomous vehicle should continue to advance and in what direction (e.g., whether the position of the wheels needs to be adjusted).

In an embodiment, the computer system uses the neural network itself to determine what data needs to be extracted from the sensor data. Specifically, the computer system retrieves a plurality of identifiers associated with a plurality of characteristics from the neural network. For example, the neural network may include an application programming interface that enables a request that the neural network will accept as input characteristics of the traffic maneuver. Each neural network (i.e., for different maneuvers) may accept different characteristics as inputs. The computer system may receive, in response to the request, an identifier of a characteristic that the neural network may accept. The identifier may simply be the name of the property. In an embodiment, each identifier may be a string of alphanumeric characters representing a particular characteristic. The autonomous vehicle may include a list (e.g., a table) that includes the identifiers and corresponding characteristics.

The computer system compares each identifier of the plurality of identifiers to an identifier of a data value in the sensor data and retrieves one or more values corresponding to the plurality of characteristics based on the comparison. For example, for maneuvers involving incorporation into an uncontrolled traffic intersection, the characteristics may include one or more identifiers of one or more other vehicles driving toward the intersection, a velocity of each of the one or more other vehicles driving toward the intersection, a distance of each of the one or more other vehicles from a particular vehicle toward the intersection, a lane of each of the one or more other vehicles driving toward the intersection, a velocity of the autonomous vehicle, a distance of the autonomous vehicle from the intersection, and other suitable characteristics.

In embodiments that include a scenario incorporating an uncontrolled traffic intersection, the computer system may receive the output of the neural network as an indication of whether the autonomous vehicle should or should not be incorporated into the intersection. The indication may be a boolean output (e.g., yes/no, 1/0, or another suitable indication). In an embodiment, in response to receiving an indication to merge into an intersection, the computer system generates one or more driving commands to operate the autonomous vehicle to merge into the intersection. In other embodiments, the neural network may output driving instructions instead of or in addition to the indication. For example, the neural network may output a heading and a rate at which to continue to advance. If the output of the neural network is to avoid merging, the output of the rate may be zero.

In an embodiment, in response to receiving an indication to avoid merging into an intersection, the computer system may execute a neural network with updated sensor data after a threshold period of time has elapsed. For example, in addition to an output indicating that the autonomous vehicle should avoid incorporation into an uncontrolled traffic intersection, the neural network may output data related to the time to wait until the neural network is executed again. This time may be based on the time required for the condition to be eliminated (e.g., another vehicle approaching the uncontrolled intersection is too close for the merge maneuver).

As discussed above, some neural networks may be configured to output one or more driving commands, and these driving commands may be used by the autonomous vehicle (e.g., by instructing the control circuitry to operate the autonomous vehicle according to the one or more driving commands) for self-operation. For example, the computer system may receive these driving commands as outputs of a neural network and send the commands to the control circuitry of the vehicle. The driving instructions may include instructions to perform a particular maneuver. The control circuitry of the vehicle may adjust the instructions (e.g., modify the instructions to adapt to the scene) based on the orientation and position of the vehicle.

The acts discussed above may be performed at an autonomous vehicle, which may include one or more computer processors and one or more non-transitory storage media storing instructions that, when executed by the one or more computer processors, cause performance of the acts described herein. Additionally or alternatively, commands for performing the actions described herein may be stored on one or more non-transitory storage media, which when executed by one or more computing devices, cause the commands to be performed.

In the previous description, embodiments of the invention have been described with reference to numerous specific details that may vary from implementation to implementation. The specification and drawings are, accordingly, to be regarded in an illustrative rather than a restrictive sense. The sole and exclusive indicator of the scope of the invention, and what is intended by the applicants to be the scope of the invention, is the literal and equivalent scope of the claims, including any subsequent correction, from the application to which the claims are entitled. Any definitions expressly set forth herein for terms contained in such claims shall govern the meaning of such terms as used in the claims. Additionally, when the term "further comprising" is used in the preceding description or the appended claims, the following of the phrase may be additional steps or entities, or sub-steps/sub-entities of previously described steps or entities.

Claims

1. A computer-implemented method, comprising:

determining, using a computer system, driving behaviors of a plurality of manually operated vehicles, each manually operated vehicle having performed a traffic merge behavior at a corresponding uncontrolled traffic intersection;

determining, with the computer system, an autonomous vehicle driving model based on the driving behavior, the autonomous vehicle driving model describing an autonomous vehicle driving behavior to be achieved by autonomous vehicle driving to merge with traffic at an uncontrolled traffic intersection; and

operating the autonomous vehicle according to the autonomous vehicle driving model using a control circuit.

2. The method of claim 1, further comprising: a tracking system is placed at or near one or more uncontrolled traffic intersections to track the driving behavior.

3. The method of claim 1, wherein determining the driving behavior comprises:

identifying vehicles that merge into respective uncontrolled traffic intersections; and

the following were determined: (1) the respective velocities of a plurality of other vehicles driving toward respective uncontrolled traffic intersections; (2) a distance of each other vehicle of the plurality of other vehicles from the merging vehicle; and (3) the lanes of each of the other vehicles.

4. The method of claim 1, wherein determining the driving behavior comprises:

identifying vehicles that do not incorporate a respective uncontrolled traffic intersection; and

for each of the vehicles that are not incorporated into the respective uncontrolled traffic intersection, a respective velocity of other vehicles in the vicinity of the each vehicle that are driving toward the respective uncontrolled traffic intersection, a distance of each of the other vehicles from the each vehicle that is not being merged, and a lane of each other vehicle are determined.

5. The method of claim 1, wherein determining the driving behavior comprises:

collecting data;

identifying a data set from the collected data; and

fitting the data set to a baseline model.

6. The method according to any one of claims 1 to 5, wherein the driving behavior comprises a speed and heading of each manually operated vehicle.

7. The method of any of claims 1-5, wherein the autonomous vehicle driving model maps, for a particular vehicle being driven toward the uncontrolled intersection, a velocity of other vehicles being driven toward the uncontrolled intersection, a distance of each other vehicle from the particular vehicle, and a lane of each other vehicle.

8. The method of any of claims 1-5, wherein the computer system, in determining the autonomous vehicle driving model, is configured to determine whether the manually operated vehicle is attempting to merge with traffic at the uncontrolled traffic intersection or is not attempting to merge with traffic at the uncontrolled traffic intersection.

9. The method of any of claims 1-5, wherein operating the autonomous vehicle according to the autonomous vehicle driving model comprises:

determining a distance from the autonomous vehicle to another vehicle driving toward the uncontrolled traffic intersection;

determining a velocity of the other vehicle;

processing, using the computer system, the distance and the velocity according to the autonomous vehicle driving model; and

providing instructions to the control circuit to merge or not merge with traffic at the uncontrolled traffic intersection based on results of processing the distance and the velocity.

10. A system for self-operation of a vehicle, comprising:

a computer system configured to determine driving behaviors of a plurality of manually operated vehicles, each manually operated vehicle having a traffic merging behavior at a respective uncontrolled traffic intersection; and

a vehicle, comprising:

one or more computer-readable media storing computer-executable instructions;

one or more processors communicatively coupled to the computer system and configured to execute the computer-executable instructions, the execution performing operations comprising:

receiving, with the vehicle, the driving behavior from the computer system;

determining, with the vehicle, an autonomous vehicle driving model based on the driving behavior, the autonomous vehicle driving model describing autonomous vehicle driving behavior to be achieved by an autonomous vehicle conducting traffic merging behavior at an uncontrolled traffic intersection; and

operating by itself with the vehicle according to the autonomous vehicle driving model.

11. The system of claim 10, wherein the computer system is separate from the vehicle.

12. The system of claim 11, wherein the computer system is configured to provide the driving behavior to the vehicle.

13. The system of claim 10, wherein the computer system is configured on the vehicle.

14. A computer-implemented method, comprising:

receiving a data set describing driving behavior of a plurality of manually operated vehicles, each manually operated vehicle attempting to perform a vehicle maneuver, wherein the data set includes a plurality of entries and a plurality of decisions taken by a human driver in performing a respective vehicle maneuver, and wherein each entry includes a plurality of fields for a plurality of factors associated with the respective vehicle maneuver;

extracting the plurality of fields from the dataset;

identifying a plurality of data types associated with the plurality of fields;

determining a plurality of characteristics associated with the traffic maneuver based on the plurality of fields;

configuring a neural network using the plurality of characteristics and the plurality of data types;

inputting data associated with the data set into the neural network in accordance with each of the plurality of characteristics; and

executing a training routine for the neural network using the input data and the plurality of characteristics associated with the traffic maneuver.

15. The method of claim 14, further comprising:

receiving sensor data from one or more sensors of an autonomous vehicle;

selecting the traffic maneuver to navigate through a portion of a trajectory based on the sensor data;

extracting one or more values corresponding to the plurality of characteristics from the sensor data;

executing the neural network with more than one of the plurality of characteristics as input data;

receiving an output from the neural network; and

operating the autonomous vehicle using control circuitry based on an output of the neural network.

16. The method of claim 15, wherein extracting the one or more values corresponding to the plurality of characteristics from the sensor data comprises:

retrieving a plurality of identifiers associated with the plurality of characteristics from the neural network;

comparing each identifier of the plurality of identifiers to an identifier of a data value in the sensor data;

retrieving the one or more values corresponding to the plurality of characteristics based on the comparison.

17. The method of claim 15, wherein the plurality of characteristics comprises:

one or more identifiers of one or more other vehicles driving toward the intersection;

a velocity of each of the one or more other vehicles driving toward the intersection;

a distance of each other vehicle of the one or more other vehicles from a particular vehicle toward the intersection;

a lane of each of the one or more other vehicles toward the intersection;

a velocity of the autonomous vehicle; and

a distance of the autonomous vehicle from the intersection.

18. The method of any one of claims 14 to 17, wherein receiving the output from the neural network comprises: receiving one or more driving commands, and wherein operating the vehicle based on the output of the neural network comprises: instructing a control circuit to operate the vehicle based on the one or more driving commands.

19. The method of any of claims 14 to 17, wherein fetching the plurality of fields from the dataset comprises:

executing a command requesting the plurality of fields using an API associated with the data set, the API being an application programming interface; and

receiving a plurality of identifiers associated with the plurality of fields in response to the command.

20. The method of any of claims 14-17, wherein identifying the plurality of data types associated with the plurality of fields comprises:

selecting, for each field of the plurality of fields, an identifier associated with the respective field;

comparing each selected identifier to a plurality of identifiers associated with known data types; and

a respective data type for each of the plurality of fields is identified based on the comparison.